{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 50,
  "global_step": 3431,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00029152394140368777,
      "grad_norm": 76.42342663863498,
      "learning_rate": 0.0,
      "loss": 2.5306,
      "step": 1
    },
    {
      "epoch": 0.0005830478828073755,
      "grad_norm": 95.87615029376128,
      "learning_rate": 9.70873786407767e-08,
      "loss": 2.5209,
      "step": 2
    },
    {
      "epoch": 0.0008745718242110633,
      "grad_norm": 78.12009399935343,
      "learning_rate": 1.941747572815534e-07,
      "loss": 2.6071,
      "step": 3
    },
    {
      "epoch": 0.001166095765614751,
      "grad_norm": 145.33518633068869,
      "learning_rate": 2.9126213592233014e-07,
      "loss": 2.4807,
      "step": 4
    },
    {
      "epoch": 0.0014576197070184389,
      "grad_norm": 83.97944277700884,
      "learning_rate": 3.883495145631068e-07,
      "loss": 2.7381,
      "step": 5
    },
    {
      "epoch": 0.0017491436484221266,
      "grad_norm": 97.29888743070265,
      "learning_rate": 4.854368932038835e-07,
      "loss": 2.3991,
      "step": 6
    },
    {
      "epoch": 0.0020406675898258144,
      "grad_norm": 65.97070089194962,
      "learning_rate": 5.825242718446603e-07,
      "loss": 2.4631,
      "step": 7
    },
    {
      "epoch": 0.002332191531229502,
      "grad_norm": 64.47891611612401,
      "learning_rate": 6.79611650485437e-07,
      "loss": 2.3181,
      "step": 8
    },
    {
      "epoch": 0.00262371547263319,
      "grad_norm": 75.50975896677284,
      "learning_rate": 7.766990291262136e-07,
      "loss": 2.5948,
      "step": 9
    },
    {
      "epoch": 0.0029152394140368777,
      "grad_norm": 106.69716542133966,
      "learning_rate": 8.737864077669904e-07,
      "loss": 2.527,
      "step": 10
    },
    {
      "epoch": 0.0032067633554405655,
      "grad_norm": 71.44707275142953,
      "learning_rate": 9.70873786407767e-07,
      "loss": 2.1419,
      "step": 11
    },
    {
      "epoch": 0.0034982872968442533,
      "grad_norm": 48.96527774305591,
      "learning_rate": 1.0679611650485437e-06,
      "loss": 2.274,
      "step": 12
    },
    {
      "epoch": 0.003789811238247941,
      "grad_norm": 55.47976120089825,
      "learning_rate": 1.1650485436893206e-06,
      "loss": 2.1934,
      "step": 13
    },
    {
      "epoch": 0.004081335179651629,
      "grad_norm": 53.92742220372767,
      "learning_rate": 1.2621359223300972e-06,
      "loss": 2.2557,
      "step": 14
    },
    {
      "epoch": 0.004372859121055317,
      "grad_norm": 39.972217603661434,
      "learning_rate": 1.359223300970874e-06,
      "loss": 1.8457,
      "step": 15
    },
    {
      "epoch": 0.004664383062459004,
      "grad_norm": 85.2173202238744,
      "learning_rate": 1.4563106796116506e-06,
      "loss": 1.9963,
      "step": 16
    },
    {
      "epoch": 0.004955907003862692,
      "grad_norm": 100.8826751210981,
      "learning_rate": 1.5533980582524272e-06,
      "loss": 1.9982,
      "step": 17
    },
    {
      "epoch": 0.00524743094526638,
      "grad_norm": 58.609884983559056,
      "learning_rate": 1.650485436893204e-06,
      "loss": 1.8656,
      "step": 18
    },
    {
      "epoch": 0.005538954886670068,
      "grad_norm": 33.075683342960154,
      "learning_rate": 1.7475728155339808e-06,
      "loss": 1.8492,
      "step": 19
    },
    {
      "epoch": 0.0058304788280737554,
      "grad_norm": 65.13568840482493,
      "learning_rate": 1.8446601941747574e-06,
      "loss": 1.9746,
      "step": 20
    },
    {
      "epoch": 0.006122002769477443,
      "grad_norm": 36.585341077750726,
      "learning_rate": 1.941747572815534e-06,
      "loss": 1.9756,
      "step": 21
    },
    {
      "epoch": 0.006413526710881131,
      "grad_norm": 42.151534100555544,
      "learning_rate": 2.0388349514563107e-06,
      "loss": 1.6761,
      "step": 22
    },
    {
      "epoch": 0.006705050652284819,
      "grad_norm": 20.68012546238523,
      "learning_rate": 2.1359223300970874e-06,
      "loss": 1.5479,
      "step": 23
    },
    {
      "epoch": 0.0069965745936885065,
      "grad_norm": 35.72432912802366,
      "learning_rate": 2.2330097087378645e-06,
      "loss": 1.4262,
      "step": 24
    },
    {
      "epoch": 0.007288098535092194,
      "grad_norm": 34.06798968201513,
      "learning_rate": 2.330097087378641e-06,
      "loss": 2.0694,
      "step": 25
    },
    {
      "epoch": 0.007579622476495882,
      "grad_norm": 20.670250660622735,
      "learning_rate": 2.427184466019418e-06,
      "loss": 1.482,
      "step": 26
    },
    {
      "epoch": 0.00787114641789957,
      "grad_norm": 73.00257914006865,
      "learning_rate": 2.5242718446601945e-06,
      "loss": 1.7858,
      "step": 27
    },
    {
      "epoch": 0.008162670359303258,
      "grad_norm": 31.399092581912143,
      "learning_rate": 2.621359223300971e-06,
      "loss": 1.7205,
      "step": 28
    },
    {
      "epoch": 0.008454194300706945,
      "grad_norm": 39.79932748372587,
      "learning_rate": 2.718446601941748e-06,
      "loss": 1.5398,
      "step": 29
    },
    {
      "epoch": 0.008745718242110633,
      "grad_norm": 37.283599177904314,
      "learning_rate": 2.8155339805825245e-06,
      "loss": 1.5772,
      "step": 30
    },
    {
      "epoch": 0.009037242183514321,
      "grad_norm": 24.3775343158831,
      "learning_rate": 2.912621359223301e-06,
      "loss": 1.3581,
      "step": 31
    },
    {
      "epoch": 0.009328766124918009,
      "grad_norm": 32.29320625203308,
      "learning_rate": 3.0097087378640778e-06,
      "loss": 1.6648,
      "step": 32
    },
    {
      "epoch": 0.009620290066321696,
      "grad_norm": 125.62914444298984,
      "learning_rate": 3.1067961165048544e-06,
      "loss": 1.5287,
      "step": 33
    },
    {
      "epoch": 0.009911814007725384,
      "grad_norm": 47.614570594174474,
      "learning_rate": 3.2038834951456315e-06,
      "loss": 1.6787,
      "step": 34
    },
    {
      "epoch": 0.010203337949129072,
      "grad_norm": 1519.7988943311027,
      "learning_rate": 3.300970873786408e-06,
      "loss": 1.7173,
      "step": 35
    },
    {
      "epoch": 0.01049486189053276,
      "grad_norm": 35.5042925851136,
      "learning_rate": 3.398058252427185e-06,
      "loss": 1.5738,
      "step": 36
    },
    {
      "epoch": 0.010786385831936448,
      "grad_norm": 21.939743293895813,
      "learning_rate": 3.4951456310679615e-06,
      "loss": 1.4179,
      "step": 37
    },
    {
      "epoch": 0.011077909773340135,
      "grad_norm": 18.992950170247735,
      "learning_rate": 3.592233009708738e-06,
      "loss": 1.5727,
      "step": 38
    },
    {
      "epoch": 0.011369433714743823,
      "grad_norm": 22.94554012015316,
      "learning_rate": 3.689320388349515e-06,
      "loss": 1.5109,
      "step": 39
    },
    {
      "epoch": 0.011660957656147511,
      "grad_norm": 104.34077412975925,
      "learning_rate": 3.7864077669902915e-06,
      "loss": 1.6397,
      "step": 40
    },
    {
      "epoch": 0.011952481597551199,
      "grad_norm": 27.629955528868642,
      "learning_rate": 3.883495145631068e-06,
      "loss": 1.5309,
      "step": 41
    },
    {
      "epoch": 0.012244005538954886,
      "grad_norm": 27.92622602176152,
      "learning_rate": 3.980582524271845e-06,
      "loss": 1.4931,
      "step": 42
    },
    {
      "epoch": 0.012535529480358574,
      "grad_norm": 73.52940364367565,
      "learning_rate": 4.0776699029126215e-06,
      "loss": 1.6857,
      "step": 43
    },
    {
      "epoch": 0.012827053421762262,
      "grad_norm": 37.659318935089765,
      "learning_rate": 4.1747572815533986e-06,
      "loss": 1.5055,
      "step": 44
    },
    {
      "epoch": 0.01311857736316595,
      "grad_norm": 40.934291529612054,
      "learning_rate": 4.271844660194175e-06,
      "loss": 1.3478,
      "step": 45
    },
    {
      "epoch": 0.013410101304569638,
      "grad_norm": 24.17557176422637,
      "learning_rate": 4.368932038834952e-06,
      "loss": 1.3864,
      "step": 46
    },
    {
      "epoch": 0.013701625245973325,
      "grad_norm": 69.5229414315914,
      "learning_rate": 4.466019417475729e-06,
      "loss": 1.6979,
      "step": 47
    },
    {
      "epoch": 0.013993149187377013,
      "grad_norm": 20.05887877339462,
      "learning_rate": 4.563106796116505e-06,
      "loss": 1.4761,
      "step": 48
    },
    {
      "epoch": 0.0142846731287807,
      "grad_norm": 50.893036882563834,
      "learning_rate": 4.660194174757282e-06,
      "loss": 1.6158,
      "step": 49
    },
    {
      "epoch": 0.014576197070184389,
      "grad_norm": 24.86701954148864,
      "learning_rate": 4.7572815533980585e-06,
      "loss": 1.5184,
      "step": 50
    },
    {
      "epoch": 0.014576197070184389,
      "eval_loss": 1.4439671039581299,
      "eval_runtime": 1947.8429,
      "eval_samples_per_second": 8.866,
      "eval_steps_per_second": 0.493,
      "step": 50
    },
    {
      "epoch": 0.014867721011588076,
      "grad_norm": 20.252266645964365,
      "learning_rate": 4.854368932038836e-06,
      "loss": 1.3417,
      "step": 51
    },
    {
      "epoch": 0.015159244952991764,
      "grad_norm": 29.318317140285615,
      "learning_rate": 4.951456310679612e-06,
      "loss": 1.2651,
      "step": 52
    },
    {
      "epoch": 0.015450768894395452,
      "grad_norm": 94.15820885885536,
      "learning_rate": 5.048543689320389e-06,
      "loss": 1.2607,
      "step": 53
    },
    {
      "epoch": 0.01574229283579914,
      "grad_norm": 53.24908659554932,
      "learning_rate": 5.145631067961165e-06,
      "loss": 1.2377,
      "step": 54
    },
    {
      "epoch": 0.01603381677720283,
      "grad_norm": 29.2695278823181,
      "learning_rate": 5.242718446601942e-06,
      "loss": 1.474,
      "step": 55
    },
    {
      "epoch": 0.016325340718606515,
      "grad_norm": 38.82049998142526,
      "learning_rate": 5.3398058252427185e-06,
      "loss": 1.4284,
      "step": 56
    },
    {
      "epoch": 0.016616864660010205,
      "grad_norm": 58.57646962903516,
      "learning_rate": 5.436893203883496e-06,
      "loss": 1.205,
      "step": 57
    },
    {
      "epoch": 0.01690838860141389,
      "grad_norm": 35.54923630963185,
      "learning_rate": 5.533980582524272e-06,
      "loss": 1.4456,
      "step": 58
    },
    {
      "epoch": 0.01719991254281758,
      "grad_norm": 64.99946344181795,
      "learning_rate": 5.631067961165049e-06,
      "loss": 1.474,
      "step": 59
    },
    {
      "epoch": 0.017491436484221266,
      "grad_norm": 26.049648537407563,
      "learning_rate": 5.728155339805825e-06,
      "loss": 1.469,
      "step": 60
    },
    {
      "epoch": 0.017782960425624956,
      "grad_norm": 28.732660454456166,
      "learning_rate": 5.825242718446602e-06,
      "loss": 1.1788,
      "step": 61
    },
    {
      "epoch": 0.018074484367028642,
      "grad_norm": 26.53217504419677,
      "learning_rate": 5.9223300970873785e-06,
      "loss": 1.1752,
      "step": 62
    },
    {
      "epoch": 0.01836600830843233,
      "grad_norm": 26.942352148232516,
      "learning_rate": 6.0194174757281556e-06,
      "loss": 1.4401,
      "step": 63
    },
    {
      "epoch": 0.018657532249836017,
      "grad_norm": 39.73836723729676,
      "learning_rate": 6.116504854368932e-06,
      "loss": 1.4821,
      "step": 64
    },
    {
      "epoch": 0.018949056191239707,
      "grad_norm": 19.07755690058887,
      "learning_rate": 6.213592233009709e-06,
      "loss": 1.3399,
      "step": 65
    },
    {
      "epoch": 0.019240580132643393,
      "grad_norm": 23.32135168698343,
      "learning_rate": 6.310679611650487e-06,
      "loss": 1.2615,
      "step": 66
    },
    {
      "epoch": 0.019532104074047082,
      "grad_norm": 44.494344426345336,
      "learning_rate": 6.407766990291263e-06,
      "loss": 1.388,
      "step": 67
    },
    {
      "epoch": 0.01982362801545077,
      "grad_norm": 31.56309648806651,
      "learning_rate": 6.50485436893204e-06,
      "loss": 1.4027,
      "step": 68
    },
    {
      "epoch": 0.020115151956854458,
      "grad_norm": 33.44891181552436,
      "learning_rate": 6.601941747572816e-06,
      "loss": 1.1468,
      "step": 69
    },
    {
      "epoch": 0.020406675898258144,
      "grad_norm": 65.89259181107087,
      "learning_rate": 6.6990291262135935e-06,
      "loss": 1.3431,
      "step": 70
    },
    {
      "epoch": 0.020698199839661834,
      "grad_norm": 61.29314176143393,
      "learning_rate": 6.79611650485437e-06,
      "loss": 1.1483,
      "step": 71
    },
    {
      "epoch": 0.02098972378106552,
      "grad_norm": 31.323614788428312,
      "learning_rate": 6.893203883495147e-06,
      "loss": 1.3256,
      "step": 72
    },
    {
      "epoch": 0.02128124772246921,
      "grad_norm": 19.449959923430875,
      "learning_rate": 6.990291262135923e-06,
      "loss": 1.2329,
      "step": 73
    },
    {
      "epoch": 0.021572771663872895,
      "grad_norm": 37.099325355767014,
      "learning_rate": 7.0873786407767e-06,
      "loss": 1.3984,
      "step": 74
    },
    {
      "epoch": 0.021864295605276585,
      "grad_norm": 270.389308250121,
      "learning_rate": 7.184466019417476e-06,
      "loss": 1.4087,
      "step": 75
    },
    {
      "epoch": 0.02215581954668027,
      "grad_norm": 17.926206039100368,
      "learning_rate": 7.2815533980582534e-06,
      "loss": 1.258,
      "step": 76
    },
    {
      "epoch": 0.02244734348808396,
      "grad_norm": 31.860030414559887,
      "learning_rate": 7.37864077669903e-06,
      "loss": 1.3007,
      "step": 77
    },
    {
      "epoch": 0.022738867429487646,
      "grad_norm": 24.749508643923,
      "learning_rate": 7.475728155339807e-06,
      "loss": 1.2126,
      "step": 78
    },
    {
      "epoch": 0.023030391370891336,
      "grad_norm": 31.849952200208822,
      "learning_rate": 7.572815533980583e-06,
      "loss": 1.2183,
      "step": 79
    },
    {
      "epoch": 0.023321915312295022,
      "grad_norm": 25.43045374149529,
      "learning_rate": 7.66990291262136e-06,
      "loss": 1.196,
      "step": 80
    },
    {
      "epoch": 0.02361343925369871,
      "grad_norm": 20.597983940843147,
      "learning_rate": 7.766990291262136e-06,
      "loss": 1.3665,
      "step": 81
    },
    {
      "epoch": 0.023904963195102397,
      "grad_norm": 26.641382204460395,
      "learning_rate": 7.864077669902913e-06,
      "loss": 1.1067,
      "step": 82
    },
    {
      "epoch": 0.024196487136506087,
      "grad_norm": 21.329017870414415,
      "learning_rate": 7.96116504854369e-06,
      "loss": 1.254,
      "step": 83
    },
    {
      "epoch": 0.024488011077909773,
      "grad_norm": 37.78375364923183,
      "learning_rate": 8.058252427184466e-06,
      "loss": 1.4461,
      "step": 84
    },
    {
      "epoch": 0.024779535019313462,
      "grad_norm": 24.612032621116708,
      "learning_rate": 8.155339805825243e-06,
      "loss": 1.3655,
      "step": 85
    },
    {
      "epoch": 0.02507105896071715,
      "grad_norm": 23.679717668011897,
      "learning_rate": 8.25242718446602e-06,
      "loss": 1.1483,
      "step": 86
    },
    {
      "epoch": 0.025362582902120838,
      "grad_norm": 24.046516641945068,
      "learning_rate": 8.349514563106797e-06,
      "loss": 1.3289,
      "step": 87
    },
    {
      "epoch": 0.025654106843524524,
      "grad_norm": 32.398999387662734,
      "learning_rate": 8.446601941747573e-06,
      "loss": 1.2696,
      "step": 88
    },
    {
      "epoch": 0.025945630784928213,
      "grad_norm": 39.78083823196173,
      "learning_rate": 8.54368932038835e-06,
      "loss": 1.2626,
      "step": 89
    },
    {
      "epoch": 0.0262371547263319,
      "grad_norm": 33.926648922572596,
      "learning_rate": 8.640776699029127e-06,
      "loss": 1.1699,
      "step": 90
    },
    {
      "epoch": 0.02652867866773559,
      "grad_norm": 65.47101570257676,
      "learning_rate": 8.737864077669904e-06,
      "loss": 1.2669,
      "step": 91
    },
    {
      "epoch": 0.026820202609139275,
      "grad_norm": 33.63276392228138,
      "learning_rate": 8.834951456310681e-06,
      "loss": 1.316,
      "step": 92
    },
    {
      "epoch": 0.027111726550542965,
      "grad_norm": 31.91334065670277,
      "learning_rate": 8.932038834951458e-06,
      "loss": 1.2806,
      "step": 93
    },
    {
      "epoch": 0.02740325049194665,
      "grad_norm": 40.68834417920064,
      "learning_rate": 9.029126213592233e-06,
      "loss": 1.3487,
      "step": 94
    },
    {
      "epoch": 0.02769477443335034,
      "grad_norm": 29.576137876597834,
      "learning_rate": 9.12621359223301e-06,
      "loss": 1.2371,
      "step": 95
    },
    {
      "epoch": 0.027986298374754026,
      "grad_norm": 124.92482530988336,
      "learning_rate": 9.223300970873788e-06,
      "loss": 1.255,
      "step": 96
    },
    {
      "epoch": 0.028277822316157716,
      "grad_norm": 28.752429756855136,
      "learning_rate": 9.320388349514565e-06,
      "loss": 1.4001,
      "step": 97
    },
    {
      "epoch": 0.0285693462575614,
      "grad_norm": 29.2641948051506,
      "learning_rate": 9.41747572815534e-06,
      "loss": 1.2938,
      "step": 98
    },
    {
      "epoch": 0.02886087019896509,
      "grad_norm": 36.81529916635092,
      "learning_rate": 9.514563106796117e-06,
      "loss": 1.3586,
      "step": 99
    },
    {
      "epoch": 0.029152394140368777,
      "grad_norm": 102.3618890161876,
      "learning_rate": 9.611650485436894e-06,
      "loss": 1.3476,
      "step": 100
    },
    {
      "epoch": 0.029152394140368777,
      "eval_loss": 1.2926723957061768,
      "eval_runtime": 1962.8181,
      "eval_samples_per_second": 8.799,
      "eval_steps_per_second": 0.489,
      "step": 100
    },
    {
      "epoch": 0.029443918081772467,
      "grad_norm": 29.71955273386497,
      "learning_rate": 9.708737864077671e-06,
      "loss": 1.1279,
      "step": 101
    },
    {
      "epoch": 0.029735442023176153,
      "grad_norm": 26.92024479421597,
      "learning_rate": 9.805825242718447e-06,
      "loss": 1.4092,
      "step": 102
    },
    {
      "epoch": 0.030026965964579842,
      "grad_norm": 33.74082530674636,
      "learning_rate": 9.902912621359224e-06,
      "loss": 1.1841,
      "step": 103
    },
    {
      "epoch": 0.03031848990598353,
      "grad_norm": 35.847309612940556,
      "learning_rate": 1e-05,
      "loss": 1.3673,
      "step": 104
    },
    {
      "epoch": 0.030610013847387218,
      "grad_norm": 38.106016039872365,
      "learning_rate": 9.99999777221597e-06,
      "loss": 1.3249,
      "step": 105
    },
    {
      "epoch": 0.030901537788790904,
      "grad_norm": 34.127918407124135,
      "learning_rate": 9.999991088865861e-06,
      "loss": 1.4068,
      "step": 106
    },
    {
      "epoch": 0.031193061730194593,
      "grad_norm": 355.8189878001581,
      "learning_rate": 9.999979949955631e-06,
      "loss": 1.164,
      "step": 107
    },
    {
      "epoch": 0.03148458567159828,
      "grad_norm": 48.901749068465,
      "learning_rate": 9.999964355495207e-06,
      "loss": 1.2437,
      "step": 108
    },
    {
      "epoch": 0.031776109613001965,
      "grad_norm": 36.34112562730177,
      "learning_rate": 9.999944305498483e-06,
      "loss": 1.159,
      "step": 109
    },
    {
      "epoch": 0.03206763355440566,
      "grad_norm": 33.00558936540292,
      "learning_rate": 9.999919799983327e-06,
      "loss": 1.0764,
      "step": 110
    },
    {
      "epoch": 0.032359157495809344,
      "grad_norm": 24.457651176422473,
      "learning_rate": 9.999890838971577e-06,
      "loss": 1.2354,
      "step": 111
    },
    {
      "epoch": 0.03265068143721303,
      "grad_norm": 27.02709798385233,
      "learning_rate": 9.99985742248904e-06,
      "loss": 1.2255,
      "step": 112
    },
    {
      "epoch": 0.03294220537861672,
      "grad_norm": 33.761689020898075,
      "learning_rate": 9.99981955056549e-06,
      "loss": 1.2119,
      "step": 113
    },
    {
      "epoch": 0.03323372932002041,
      "grad_norm": 33.77625426050327,
      "learning_rate": 9.999777223234682e-06,
      "loss": 1.1886,
      "step": 114
    },
    {
      "epoch": 0.033525253261424096,
      "grad_norm": 34.99790265671646,
      "learning_rate": 9.99973044053433e-06,
      "loss": 1.1547,
      "step": 115
    },
    {
      "epoch": 0.03381677720282778,
      "grad_norm": 31.662627217957038,
      "learning_rate": 9.999679202506126e-06,
      "loss": 1.3047,
      "step": 116
    },
    {
      "epoch": 0.03410830114423147,
      "grad_norm": 19.63264952763771,
      "learning_rate": 9.999623509195724e-06,
      "loss": 1.2557,
      "step": 117
    },
    {
      "epoch": 0.03439982508563516,
      "grad_norm": 45.20286908810948,
      "learning_rate": 9.999563360652757e-06,
      "loss": 1.1861,
      "step": 118
    },
    {
      "epoch": 0.03469134902703885,
      "grad_norm": 85.09155031769129,
      "learning_rate": 9.999498756930823e-06,
      "loss": 1.4304,
      "step": 119
    },
    {
      "epoch": 0.03498287296844253,
      "grad_norm": 53.72371009802662,
      "learning_rate": 9.999429698087491e-06,
      "loss": 1.3538,
      "step": 120
    },
    {
      "epoch": 0.03527439690984622,
      "grad_norm": 25.59096027880631,
      "learning_rate": 9.9993561841843e-06,
      "loss": 1.3918,
      "step": 121
    },
    {
      "epoch": 0.03556592085124991,
      "grad_norm": 44.22343813269323,
      "learning_rate": 9.99927821528676e-06,
      "loss": 1.0796,
      "step": 122
    },
    {
      "epoch": 0.0358574447926536,
      "grad_norm": 24.197305993678725,
      "learning_rate": 9.99919579146435e-06,
      "loss": 1.3199,
      "step": 123
    },
    {
      "epoch": 0.036148968734057284,
      "grad_norm": 26.01617016467829,
      "learning_rate": 9.999108912790521e-06,
      "loss": 1.3659,
      "step": 124
    },
    {
      "epoch": 0.03644049267546097,
      "grad_norm": 19.761778037072567,
      "learning_rate": 9.999017579342687e-06,
      "loss": 1.2352,
      "step": 125
    },
    {
      "epoch": 0.03673201661686466,
      "grad_norm": 40.94400448810034,
      "learning_rate": 9.99892179120224e-06,
      "loss": 1.3459,
      "step": 126
    },
    {
      "epoch": 0.03702354055826835,
      "grad_norm": 55.61535873078524,
      "learning_rate": 9.998821548454539e-06,
      "loss": 1.163,
      "step": 127
    },
    {
      "epoch": 0.037315064499672035,
      "grad_norm": 67.66187825424119,
      "learning_rate": 9.99871685118891e-06,
      "loss": 1.2222,
      "step": 128
    },
    {
      "epoch": 0.03760658844107572,
      "grad_norm": 78.4667440019241,
      "learning_rate": 9.998607699498648e-06,
      "loss": 1.3658,
      "step": 129
    },
    {
      "epoch": 0.037898112382479414,
      "grad_norm": 26.5333172574931,
      "learning_rate": 9.998494093481022e-06,
      "loss": 1.1761,
      "step": 130
    },
    {
      "epoch": 0.0381896363238831,
      "grad_norm": 32.268047230303,
      "learning_rate": 9.998376033237268e-06,
      "loss": 1.0932,
      "step": 131
    },
    {
      "epoch": 0.038481160265286786,
      "grad_norm": 63.23123972601051,
      "learning_rate": 9.998253518872592e-06,
      "loss": 1.2629,
      "step": 132
    },
    {
      "epoch": 0.03877268420669047,
      "grad_norm": 31.313629892511806,
      "learning_rate": 9.998126550496165e-06,
      "loss": 1.1769,
      "step": 133
    },
    {
      "epoch": 0.039064208148094165,
      "grad_norm": 28.174414369292936,
      "learning_rate": 9.997995128221131e-06,
      "loss": 1.2604,
      "step": 134
    },
    {
      "epoch": 0.03935573208949785,
      "grad_norm": 33.58393416102376,
      "learning_rate": 9.997859252164606e-06,
      "loss": 1.1721,
      "step": 135
    },
    {
      "epoch": 0.03964725603090154,
      "grad_norm": 24.253683175766465,
      "learning_rate": 9.997718922447669e-06,
      "loss": 1.2812,
      "step": 136
    },
    {
      "epoch": 0.03993877997230522,
      "grad_norm": 30.25223280228506,
      "learning_rate": 9.997574139195367e-06,
      "loss": 1.1007,
      "step": 137
    },
    {
      "epoch": 0.040230303913708916,
      "grad_norm": 73.5957748057368,
      "learning_rate": 9.99742490253672e-06,
      "loss": 1.3036,
      "step": 138
    },
    {
      "epoch": 0.0405218278551126,
      "grad_norm": 41.26881529885212,
      "learning_rate": 9.997271212604715e-06,
      "loss": 1.2049,
      "step": 139
    },
    {
      "epoch": 0.04081335179651629,
      "grad_norm": 26.766612803527998,
      "learning_rate": 9.99711306953631e-06,
      "loss": 1.1823,
      "step": 140
    },
    {
      "epoch": 0.041104875737919974,
      "grad_norm": 75.23390275609594,
      "learning_rate": 9.996950473472423e-06,
      "loss": 1.1875,
      "step": 141
    },
    {
      "epoch": 0.04139639967932367,
      "grad_norm": 17.709560859515793,
      "learning_rate": 9.99678342455795e-06,
      "loss": 1.0265,
      "step": 142
    },
    {
      "epoch": 0.04168792362072735,
      "grad_norm": 30.579577684333902,
      "learning_rate": 9.996611922941748e-06,
      "loss": 1.2225,
      "step": 143
    },
    {
      "epoch": 0.04197944756213104,
      "grad_norm": 39.2251222230853,
      "learning_rate": 9.996435968776646e-06,
      "loss": 1.0218,
      "step": 144
    },
    {
      "epoch": 0.042270971503534725,
      "grad_norm": 28.68518225445654,
      "learning_rate": 9.99625556221944e-06,
      "loss": 1.2246,
      "step": 145
    },
    {
      "epoch": 0.04256249544493842,
      "grad_norm": 27.8882972045983,
      "learning_rate": 9.996070703430888e-06,
      "loss": 1.1603,
      "step": 146
    },
    {
      "epoch": 0.042854019386342104,
      "grad_norm": 29.573800628837542,
      "learning_rate": 9.995881392575725e-06,
      "loss": 1.2812,
      "step": 147
    },
    {
      "epoch": 0.04314554332774579,
      "grad_norm": 19.84011564379669,
      "learning_rate": 9.995687629822647e-06,
      "loss": 1.1666,
      "step": 148
    },
    {
      "epoch": 0.043437067269149476,
      "grad_norm": 19.305772227556034,
      "learning_rate": 9.995489415344318e-06,
      "loss": 1.4562,
      "step": 149
    },
    {
      "epoch": 0.04372859121055317,
      "grad_norm": 70.33308301179466,
      "learning_rate": 9.99528674931737e-06,
      "loss": 1.2256,
      "step": 150
    },
    {
      "epoch": 0.04372859121055317,
      "eval_loss": 1.2519495487213135,
      "eval_runtime": 1925.0638,
      "eval_samples_per_second": 8.971,
      "eval_steps_per_second": 0.499,
      "step": 150
    },
    {
      "epoch": 0.044020115151956855,
      "grad_norm": 35.36816351738553,
      "learning_rate": 9.995079631922404e-06,
      "loss": 1.3255,
      "step": 151
    },
    {
      "epoch": 0.04431163909336054,
      "grad_norm": 31.963140458639014,
      "learning_rate": 9.99486806334398e-06,
      "loss": 1.1092,
      "step": 152
    },
    {
      "epoch": 0.04460316303476423,
      "grad_norm": 21.89474661413012,
      "learning_rate": 9.994652043770634e-06,
      "loss": 1.1792,
      "step": 153
    },
    {
      "epoch": 0.04489468697616792,
      "grad_norm": 29.59889787987548,
      "learning_rate": 9.994431573394861e-06,
      "loss": 1.1238,
      "step": 154
    },
    {
      "epoch": 0.045186210917571606,
      "grad_norm": 29.280944662365417,
      "learning_rate": 9.994206652413128e-06,
      "loss": 0.9367,
      "step": 155
    },
    {
      "epoch": 0.04547773485897529,
      "grad_norm": 22.56996569526786,
      "learning_rate": 9.993977281025862e-06,
      "loss": 1.3028,
      "step": 156
    },
    {
      "epoch": 0.04576925880037898,
      "grad_norm": 29.93631980199539,
      "learning_rate": 9.993743459437462e-06,
      "loss": 1.1235,
      "step": 157
    },
    {
      "epoch": 0.04606078274178267,
      "grad_norm": 82.28728253140363,
      "learning_rate": 9.993505187856289e-06,
      "loss": 1.1516,
      "step": 158
    },
    {
      "epoch": 0.04635230668318636,
      "grad_norm": 36.21037107422895,
      "learning_rate": 9.993262466494667e-06,
      "loss": 1.3255,
      "step": 159
    },
    {
      "epoch": 0.046643830624590044,
      "grad_norm": 41.22995050962949,
      "learning_rate": 9.993015295568893e-06,
      "loss": 1.1345,
      "step": 160
    },
    {
      "epoch": 0.04693535456599373,
      "grad_norm": 53.0249391878922,
      "learning_rate": 9.99276367529922e-06,
      "loss": 1.2059,
      "step": 161
    },
    {
      "epoch": 0.04722687850739742,
      "grad_norm": 25.895914314713362,
      "learning_rate": 9.992507605909873e-06,
      "loss": 1.2241,
      "step": 162
    },
    {
      "epoch": 0.04751840244880111,
      "grad_norm": 61.407427813561576,
      "learning_rate": 9.992247087629037e-06,
      "loss": 1.0531,
      "step": 163
    },
    {
      "epoch": 0.047809926390204795,
      "grad_norm": 19.304099452834844,
      "learning_rate": 9.991982120688865e-06,
      "loss": 1.2372,
      "step": 164
    },
    {
      "epoch": 0.04810145033160848,
      "grad_norm": 26.19543416785236,
      "learning_rate": 9.991712705325474e-06,
      "loss": 1.0862,
      "step": 165
    },
    {
      "epoch": 0.048392974273012174,
      "grad_norm": 78.1152343401651,
      "learning_rate": 9.99143884177894e-06,
      "loss": 1.2031,
      "step": 166
    },
    {
      "epoch": 0.04868449821441586,
      "grad_norm": 62.86581849203058,
      "learning_rate": 9.991160530293309e-06,
      "loss": 1.0308,
      "step": 167
    },
    {
      "epoch": 0.048976022155819546,
      "grad_norm": 31.91891814955325,
      "learning_rate": 9.990877771116588e-06,
      "loss": 1.2142,
      "step": 168
    },
    {
      "epoch": 0.04926754609722323,
      "grad_norm": 17.922156991191123,
      "learning_rate": 9.990590564500745e-06,
      "loss": 1.3636,
      "step": 169
    },
    {
      "epoch": 0.049559070038626925,
      "grad_norm": 14.740154207900678,
      "learning_rate": 9.99029891070172e-06,
      "loss": 1.1355,
      "step": 170
    },
    {
      "epoch": 0.04985059398003061,
      "grad_norm": 21.755293636386696,
      "learning_rate": 9.990002809979403e-06,
      "loss": 1.0605,
      "step": 171
    },
    {
      "epoch": 0.0501421179214343,
      "grad_norm": 30.27550510680315,
      "learning_rate": 9.989702262597656e-06,
      "loss": 1.3239,
      "step": 172
    },
    {
      "epoch": 0.05043364186283798,
      "grad_norm": 32.068155604514764,
      "learning_rate": 9.9893972688243e-06,
      "loss": 1.2464,
      "step": 173
    },
    {
      "epoch": 0.050725165804241676,
      "grad_norm": 18.916253444278766,
      "learning_rate": 9.989087828931121e-06,
      "loss": 1.2469,
      "step": 174
    },
    {
      "epoch": 0.05101668974564536,
      "grad_norm": 26.827195143504046,
      "learning_rate": 9.988773943193863e-06,
      "loss": 1.0847,
      "step": 175
    },
    {
      "epoch": 0.05130821368704905,
      "grad_norm": 29.554926019257167,
      "learning_rate": 9.988455611892237e-06,
      "loss": 1.1882,
      "step": 176
    },
    {
      "epoch": 0.051599737628452734,
      "grad_norm": 20.001183933438966,
      "learning_rate": 9.988132835309907e-06,
      "loss": 1.173,
      "step": 177
    },
    {
      "epoch": 0.05189126156985643,
      "grad_norm": 37.76471147118426,
      "learning_rate": 9.987805613734508e-06,
      "loss": 1.2048,
      "step": 178
    },
    {
      "epoch": 0.05218278551126011,
      "grad_norm": 17.339530320108658,
      "learning_rate": 9.987473947457631e-06,
      "loss": 1.1916,
      "step": 179
    },
    {
      "epoch": 0.0524743094526638,
      "grad_norm": 33.92094177377764,
      "learning_rate": 9.987137836774827e-06,
      "loss": 1.0881,
      "step": 180
    },
    {
      "epoch": 0.052765833394067485,
      "grad_norm": 16.378188506495995,
      "learning_rate": 9.986797281985609e-06,
      "loss": 1.2235,
      "step": 181
    },
    {
      "epoch": 0.05305735733547118,
      "grad_norm": 38.23890518261313,
      "learning_rate": 9.986452283393452e-06,
      "loss": 1.3918,
      "step": 182
    },
    {
      "epoch": 0.053348881276874864,
      "grad_norm": 32.838792384325465,
      "learning_rate": 9.986102841305788e-06,
      "loss": 1.2214,
      "step": 183
    },
    {
      "epoch": 0.05364040521827855,
      "grad_norm": 24.64027625659323,
      "learning_rate": 9.985748956034007e-06,
      "loss": 1.2001,
      "step": 184
    },
    {
      "epoch": 0.053931929159682236,
      "grad_norm": 17.197327906109653,
      "learning_rate": 9.985390627893462e-06,
      "loss": 1.0878,
      "step": 185
    },
    {
      "epoch": 0.05422345310108593,
      "grad_norm": 46.84763905233915,
      "learning_rate": 9.985027857203469e-06,
      "loss": 1.2305,
      "step": 186
    },
    {
      "epoch": 0.054514977042489615,
      "grad_norm": 29.40516212755096,
      "learning_rate": 9.984660644287292e-06,
      "loss": 1.2176,
      "step": 187
    },
    {
      "epoch": 0.0548065009838933,
      "grad_norm": 28.830043245563925,
      "learning_rate": 9.984288989472162e-06,
      "loss": 1.1284,
      "step": 188
    },
    {
      "epoch": 0.05509802492529699,
      "grad_norm": 20.32400531959014,
      "learning_rate": 9.983912893089263e-06,
      "loss": 1.1092,
      "step": 189
    },
    {
      "epoch": 0.05538954886670068,
      "grad_norm": 52.184638465326245,
      "learning_rate": 9.983532355473744e-06,
      "loss": 1.1048,
      "step": 190
    },
    {
      "epoch": 0.055681072808104366,
      "grad_norm": 23.55263787101191,
      "learning_rate": 9.983147376964703e-06,
      "loss": 1.1287,
      "step": 191
    },
    {
      "epoch": 0.05597259674950805,
      "grad_norm": 51.60523180422543,
      "learning_rate": 9.982757957905204e-06,
      "loss": 1.2864,
      "step": 192
    },
    {
      "epoch": 0.05626412069091174,
      "grad_norm": 32.346912811752816,
      "learning_rate": 9.982364098642261e-06,
      "loss": 1.0729,
      "step": 193
    },
    {
      "epoch": 0.05655564463231543,
      "grad_norm": 25.12306590805981,
      "learning_rate": 9.981965799526846e-06,
      "loss": 1.2492,
      "step": 194
    },
    {
      "epoch": 0.05684716857371912,
      "grad_norm": 29.43811443556616,
      "learning_rate": 9.98156306091389e-06,
      "loss": 1.1655,
      "step": 195
    },
    {
      "epoch": 0.0571386925151228,
      "grad_norm": 22.01306166222628,
      "learning_rate": 9.981155883162281e-06,
      "loss": 1.0572,
      "step": 196
    },
    {
      "epoch": 0.05743021645652649,
      "grad_norm": 70.73934999855636,
      "learning_rate": 9.980744266634858e-06,
      "loss": 1.2877,
      "step": 197
    },
    {
      "epoch": 0.05772174039793018,
      "grad_norm": 39.92046407367053,
      "learning_rate": 9.980328211698418e-06,
      "loss": 1.2081,
      "step": 198
    },
    {
      "epoch": 0.05801326433933387,
      "grad_norm": 26.143371346342654,
      "learning_rate": 9.979907718723715e-06,
      "loss": 1.0515,
      "step": 199
    },
    {
      "epoch": 0.058304788280737554,
      "grad_norm": 20.14183571582136,
      "learning_rate": 9.979482788085455e-06,
      "loss": 1.1113,
      "step": 200
    },
    {
      "epoch": 0.058304788280737554,
      "eval_loss": 1.230379343032837,
      "eval_runtime": 1936.0858,
      "eval_samples_per_second": 8.92,
      "eval_steps_per_second": 0.496,
      "step": 200
    },
    {
      "epoch": 0.05859631222214124,
      "grad_norm": 26.338950029629714,
      "learning_rate": 9.9790534201623e-06,
      "loss": 1.1994,
      "step": 201
    },
    {
      "epoch": 0.058887836163544933,
      "grad_norm": 75.77636235966763,
      "learning_rate": 9.978619615336866e-06,
      "loss": 1.1396,
      "step": 202
    },
    {
      "epoch": 0.05917936010494862,
      "grad_norm": 19.195665907708683,
      "learning_rate": 9.978181373995719e-06,
      "loss": 1.0422,
      "step": 203
    },
    {
      "epoch": 0.059470884046352306,
      "grad_norm": 38.88340383795867,
      "learning_rate": 9.977738696529387e-06,
      "loss": 1.1223,
      "step": 204
    },
    {
      "epoch": 0.05976240798775599,
      "grad_norm": 32.93027743750072,
      "learning_rate": 9.977291583332342e-06,
      "loss": 1.0942,
      "step": 205
    },
    {
      "epoch": 0.060053931929159685,
      "grad_norm": 62.93619473675984,
      "learning_rate": 9.976840034803014e-06,
      "loss": 1.1966,
      "step": 206
    },
    {
      "epoch": 0.06034545587056337,
      "grad_norm": 65.95146186688507,
      "learning_rate": 9.976384051343784e-06,
      "loss": 1.0689,
      "step": 207
    },
    {
      "epoch": 0.06063697981196706,
      "grad_norm": 54.64604568609321,
      "learning_rate": 9.975923633360985e-06,
      "loss": 1.2578,
      "step": 208
    },
    {
      "epoch": 0.06092850375337074,
      "grad_norm": 25.804786680825238,
      "learning_rate": 9.975458781264901e-06,
      "loss": 1.2428,
      "step": 209
    },
    {
      "epoch": 0.061220027694774436,
      "grad_norm": 23.141310346078324,
      "learning_rate": 9.974989495469771e-06,
      "loss": 1.3496,
      "step": 210
    },
    {
      "epoch": 0.06151155163617812,
      "grad_norm": 74.75269709255834,
      "learning_rate": 9.974515776393778e-06,
      "loss": 1.2449,
      "step": 211
    },
    {
      "epoch": 0.06180307557758181,
      "grad_norm": 33.34481586117105,
      "learning_rate": 9.974037624459063e-06,
      "loss": 1.1077,
      "step": 212
    },
    {
      "epoch": 0.062094599518985494,
      "grad_norm": 59.38426608558181,
      "learning_rate": 9.973555040091712e-06,
      "loss": 1.1492,
      "step": 213
    },
    {
      "epoch": 0.06238612346038919,
      "grad_norm": 32.5839174379988,
      "learning_rate": 9.973068023721761e-06,
      "loss": 1.0635,
      "step": 214
    },
    {
      "epoch": 0.06267764740179287,
      "grad_norm": 17.18132069376527,
      "learning_rate": 9.9725765757832e-06,
      "loss": 1.1484,
      "step": 215
    },
    {
      "epoch": 0.06296917134319656,
      "grad_norm": 29.4890840178089,
      "learning_rate": 9.972080696713962e-06,
      "loss": 1.1871,
      "step": 216
    },
    {
      "epoch": 0.06326069528460025,
      "grad_norm": 30.131049043670135,
      "learning_rate": 9.971580386955933e-06,
      "loss": 1.2249,
      "step": 217
    },
    {
      "epoch": 0.06355221922600393,
      "grad_norm": 82.80946274961386,
      "learning_rate": 9.971075646954946e-06,
      "loss": 1.0398,
      "step": 218
    },
    {
      "epoch": 0.06384374316740762,
      "grad_norm": 34.44336816044852,
      "learning_rate": 9.970566477160782e-06,
      "loss": 1.0038,
      "step": 219
    },
    {
      "epoch": 0.06413526710881132,
      "grad_norm": 26.75431065575965,
      "learning_rate": 9.970052878027169e-06,
      "loss": 1.2875,
      "step": 220
    },
    {
      "epoch": 0.064426791050215,
      "grad_norm": 25.027457606555686,
      "learning_rate": 9.969534850011782e-06,
      "loss": 1.1564,
      "step": 221
    },
    {
      "epoch": 0.06471831499161869,
      "grad_norm": 44.4261233477415,
      "learning_rate": 9.969012393576241e-06,
      "loss": 1.1188,
      "step": 222
    },
    {
      "epoch": 0.06500983893302237,
      "grad_norm": 23.99917930782375,
      "learning_rate": 9.968485509186116e-06,
      "loss": 1.1528,
      "step": 223
    },
    {
      "epoch": 0.06530136287442606,
      "grad_norm": 40.43556728517224,
      "learning_rate": 9.967954197310922e-06,
      "loss": 1.0944,
      "step": 224
    },
    {
      "epoch": 0.06559288681582975,
      "grad_norm": 32.930652757599276,
      "learning_rate": 9.967418458424116e-06,
      "loss": 1.1644,
      "step": 225
    },
    {
      "epoch": 0.06588441075723343,
      "grad_norm": 54.75784101350894,
      "learning_rate": 9.966878293003102e-06,
      "loss": 1.1834,
      "step": 226
    },
    {
      "epoch": 0.06617593469863713,
      "grad_norm": 76.1645542380386,
      "learning_rate": 9.966333701529231e-06,
      "loss": 1.1204,
      "step": 227
    },
    {
      "epoch": 0.06646745864004082,
      "grad_norm": 23.970560321995137,
      "learning_rate": 9.965784684487794e-06,
      "loss": 1.3227,
      "step": 228
    },
    {
      "epoch": 0.0667589825814445,
      "grad_norm": 16.832318755743298,
      "learning_rate": 9.96523124236803e-06,
      "loss": 1.0917,
      "step": 229
    },
    {
      "epoch": 0.06705050652284819,
      "grad_norm": 23.776324237313617,
      "learning_rate": 9.964673375663114e-06,
      "loss": 1.2957,
      "step": 230
    },
    {
      "epoch": 0.06734203046425187,
      "grad_norm": 40.767555946674925,
      "learning_rate": 9.964111084870174e-06,
      "loss": 1.2536,
      "step": 231
    },
    {
      "epoch": 0.06763355440565556,
      "grad_norm": 29.947818920177635,
      "learning_rate": 9.96354437049027e-06,
      "loss": 1.253,
      "step": 232
    },
    {
      "epoch": 0.06792507834705926,
      "grad_norm": 51.389163012953475,
      "learning_rate": 9.962973233028415e-06,
      "loss": 1.0496,
      "step": 233
    },
    {
      "epoch": 0.06821660228846294,
      "grad_norm": 45.810389385395084,
      "learning_rate": 9.962397672993552e-06,
      "loss": 1.1641,
      "step": 234
    },
    {
      "epoch": 0.06850812622986663,
      "grad_norm": 23.03190988264714,
      "learning_rate": 9.961817690898571e-06,
      "loss": 1.3434,
      "step": 235
    },
    {
      "epoch": 0.06879965017127032,
      "grad_norm": 24.41766051972087,
      "learning_rate": 9.961233287260305e-06,
      "loss": 1.2207,
      "step": 236
    },
    {
      "epoch": 0.069091174112674,
      "grad_norm": 21.651072136286412,
      "learning_rate": 9.96064446259952e-06,
      "loss": 1.2362,
      "step": 237
    },
    {
      "epoch": 0.0693826980540777,
      "grad_norm": 24.082432379780457,
      "learning_rate": 9.96005121744093e-06,
      "loss": 1.1387,
      "step": 238
    },
    {
      "epoch": 0.06967422199548137,
      "grad_norm": 268.81459000348866,
      "learning_rate": 9.959453552313181e-06,
      "loss": 1.0757,
      "step": 239
    },
    {
      "epoch": 0.06996574593688507,
      "grad_norm": 36.65665503025889,
      "learning_rate": 9.958851467748863e-06,
      "loss": 1.1036,
      "step": 240
    },
    {
      "epoch": 0.07025726987828876,
      "grad_norm": 22.02503034867956,
      "learning_rate": 9.958244964284497e-06,
      "loss": 1.241,
      "step": 241
    },
    {
      "epoch": 0.07054879381969244,
      "grad_norm": 24.528350717993323,
      "learning_rate": 9.957634042460551e-06,
      "loss": 1.0161,
      "step": 242
    },
    {
      "epoch": 0.07084031776109613,
      "grad_norm": 22.732945969576082,
      "learning_rate": 9.957018702821425e-06,
      "loss": 1.0719,
      "step": 243
    },
    {
      "epoch": 0.07113184170249982,
      "grad_norm": 32.220422411129064,
      "learning_rate": 9.956398945915455e-06,
      "loss": 1.1691,
      "step": 244
    },
    {
      "epoch": 0.0714233656439035,
      "grad_norm": 16.59761998988499,
      "learning_rate": 9.955774772294915e-06,
      "loss": 1.1186,
      "step": 245
    },
    {
      "epoch": 0.0717148895853072,
      "grad_norm": 37.40532109882256,
      "learning_rate": 9.955146182516015e-06,
      "loss": 1.2017,
      "step": 246
    },
    {
      "epoch": 0.07200641352671087,
      "grad_norm": 16.739065021549845,
      "learning_rate": 9.9545131771389e-06,
      "loss": 1.2893,
      "step": 247
    },
    {
      "epoch": 0.07229793746811457,
      "grad_norm": 60.45259283269908,
      "learning_rate": 9.95387575672765e-06,
      "loss": 1.2473,
      "step": 248
    },
    {
      "epoch": 0.07258946140951826,
      "grad_norm": 68.35933276544733,
      "learning_rate": 9.95323392185028e-06,
      "loss": 1.3837,
      "step": 249
    },
    {
      "epoch": 0.07288098535092194,
      "grad_norm": 45.22189820233339,
      "learning_rate": 9.952587673078738e-06,
      "loss": 1.2667,
      "step": 250
    },
    {
      "epoch": 0.07288098535092194,
      "eval_loss": 1.224450945854187,
      "eval_runtime": 1944.0493,
      "eval_samples_per_second": 8.884,
      "eval_steps_per_second": 0.494,
      "step": 250
    },
    {
      "epoch": 0.07317250929232563,
      "grad_norm": 20.99934319618223,
      "learning_rate": 9.9519370109889e-06,
      "loss": 1.1208,
      "step": 251
    },
    {
      "epoch": 0.07346403323372933,
      "grad_norm": 34.18336227600456,
      "learning_rate": 9.951281936160587e-06,
      "loss": 1.1068,
      "step": 252
    },
    {
      "epoch": 0.073755557175133,
      "grad_norm": 18.91030376599453,
      "learning_rate": 9.950622449177538e-06,
      "loss": 1.1393,
      "step": 253
    },
    {
      "epoch": 0.0740470811165367,
      "grad_norm": 21.204253265962823,
      "learning_rate": 9.949958550627436e-06,
      "loss": 0.9888,
      "step": 254
    },
    {
      "epoch": 0.07433860505794038,
      "grad_norm": 36.138393589798675,
      "learning_rate": 9.94929024110189e-06,
      "loss": 1.1287,
      "step": 255
    },
    {
      "epoch": 0.07463012899934407,
      "grad_norm": 22.548122365383733,
      "learning_rate": 9.948617521196438e-06,
      "loss": 1.2028,
      "step": 256
    },
    {
      "epoch": 0.07492165294074776,
      "grad_norm": 28.94037442097268,
      "learning_rate": 9.947940391510549e-06,
      "loss": 1.1777,
      "step": 257
    },
    {
      "epoch": 0.07521317688215144,
      "grad_norm": 99.98720468370647,
      "learning_rate": 9.947258852647623e-06,
      "loss": 1.1966,
      "step": 258
    },
    {
      "epoch": 0.07550470082355513,
      "grad_norm": 221.2581265025932,
      "learning_rate": 9.946572905214989e-06,
      "loss": 1.1616,
      "step": 259
    },
    {
      "epoch": 0.07579622476495883,
      "grad_norm": 38.98670689998384,
      "learning_rate": 9.945882549823906e-06,
      "loss": 1.1117,
      "step": 260
    },
    {
      "epoch": 0.0760877487063625,
      "grad_norm": 20.962674271304802,
      "learning_rate": 9.945187787089556e-06,
      "loss": 1.0036,
      "step": 261
    },
    {
      "epoch": 0.0763792726477662,
      "grad_norm": 29.861070945767604,
      "learning_rate": 9.944488617631053e-06,
      "loss": 1.2921,
      "step": 262
    },
    {
      "epoch": 0.07667079658916988,
      "grad_norm": 28.38138460113728,
      "learning_rate": 9.943785042071437e-06,
      "loss": 0.9902,
      "step": 263
    },
    {
      "epoch": 0.07696232053057357,
      "grad_norm": 37.826643063107475,
      "learning_rate": 9.943077061037672e-06,
      "loss": 1.2201,
      "step": 264
    },
    {
      "epoch": 0.07725384447197726,
      "grad_norm": 26.250655516355973,
      "learning_rate": 9.942364675160651e-06,
      "loss": 1.1872,
      "step": 265
    },
    {
      "epoch": 0.07754536841338094,
      "grad_norm": 24.55160743062673,
      "learning_rate": 9.94164788507519e-06,
      "loss": 1.1019,
      "step": 266
    },
    {
      "epoch": 0.07783689235478464,
      "grad_norm": 37.40364862275037,
      "learning_rate": 9.940926691420031e-06,
      "loss": 1.2374,
      "step": 267
    },
    {
      "epoch": 0.07812841629618833,
      "grad_norm": 38.26886519439301,
      "learning_rate": 9.940201094837838e-06,
      "loss": 1.1108,
      "step": 268
    },
    {
      "epoch": 0.07841994023759201,
      "grad_norm": 34.211543829343555,
      "learning_rate": 9.939471095975204e-06,
      "loss": 1.0459,
      "step": 269
    },
    {
      "epoch": 0.0787114641789957,
      "grad_norm": 36.75385899947443,
      "learning_rate": 9.938736695482636e-06,
      "loss": 1.1333,
      "step": 270
    },
    {
      "epoch": 0.07900298812039938,
      "grad_norm": 33.62430589352652,
      "learning_rate": 9.93799789401457e-06,
      "loss": 1.2609,
      "step": 271
    },
    {
      "epoch": 0.07929451206180307,
      "grad_norm": 64.74037209851723,
      "learning_rate": 9.937254692229363e-06,
      "loss": 1.3584,
      "step": 272
    },
    {
      "epoch": 0.07958603600320677,
      "grad_norm": 26.00456508571736,
      "learning_rate": 9.936507090789294e-06,
      "loss": 0.995,
      "step": 273
    },
    {
      "epoch": 0.07987755994461045,
      "grad_norm": 61.989786086293236,
      "learning_rate": 9.935755090360554e-06,
      "loss": 1.2427,
      "step": 274
    },
    {
      "epoch": 0.08016908388601414,
      "grad_norm": 33.300077819782246,
      "learning_rate": 9.934998691613269e-06,
      "loss": 1.1422,
      "step": 275
    },
    {
      "epoch": 0.08046060782741783,
      "grad_norm": 27.245759132269217,
      "learning_rate": 9.93423789522147e-06,
      "loss": 1.0779,
      "step": 276
    },
    {
      "epoch": 0.08075213176882151,
      "grad_norm": 30.603536388436044,
      "learning_rate": 9.933472701863117e-06,
      "loss": 1.1847,
      "step": 277
    },
    {
      "epoch": 0.0810436557102252,
      "grad_norm": 23.91367092785263,
      "learning_rate": 9.932703112220084e-06,
      "loss": 1.3418,
      "step": 278
    },
    {
      "epoch": 0.08133517965162888,
      "grad_norm": 20.238289247951517,
      "learning_rate": 9.931929126978158e-06,
      "loss": 1.1073,
      "step": 279
    },
    {
      "epoch": 0.08162670359303258,
      "grad_norm": 41.56929168907103,
      "learning_rate": 9.931150746827055e-06,
      "loss": 1.2071,
      "step": 280
    },
    {
      "epoch": 0.08191822753443627,
      "grad_norm": 48.71749507042547,
      "learning_rate": 9.930367972460392e-06,
      "loss": 1.011,
      "step": 281
    },
    {
      "epoch": 0.08220975147583995,
      "grad_norm": 43.527465464370295,
      "learning_rate": 9.929580804575718e-06,
      "loss": 1.0256,
      "step": 282
    },
    {
      "epoch": 0.08250127541724364,
      "grad_norm": 35.632875152751126,
      "learning_rate": 9.928789243874486e-06,
      "loss": 1.1297,
      "step": 283
    },
    {
      "epoch": 0.08279279935864733,
      "grad_norm": 16.994125759182598,
      "learning_rate": 9.927993291062064e-06,
      "loss": 0.988,
      "step": 284
    },
    {
      "epoch": 0.08308432330005101,
      "grad_norm": 40.91298504249289,
      "learning_rate": 9.927192946847739e-06,
      "loss": 1.2528,
      "step": 285
    },
    {
      "epoch": 0.0833758472414547,
      "grad_norm": 37.13969806792163,
      "learning_rate": 9.926388211944707e-06,
      "loss": 1.1821,
      "step": 286
    },
    {
      "epoch": 0.08366737118285839,
      "grad_norm": 26.19841399644935,
      "learning_rate": 9.92557908707008e-06,
      "loss": 1.4307,
      "step": 287
    },
    {
      "epoch": 0.08395889512426208,
      "grad_norm": 31.47670466705893,
      "learning_rate": 9.924765572944879e-06,
      "loss": 1.1854,
      "step": 288
    },
    {
      "epoch": 0.08425041906566577,
      "grad_norm": 22.927217657019586,
      "learning_rate": 9.92394767029404e-06,
      "loss": 1.0915,
      "step": 289
    },
    {
      "epoch": 0.08454194300706945,
      "grad_norm": 21.671809344004465,
      "learning_rate": 9.9231253798464e-06,
      "loss": 1.2148,
      "step": 290
    },
    {
      "epoch": 0.08483346694847314,
      "grad_norm": 38.00903608190354,
      "learning_rate": 9.922298702334723e-06,
      "loss": 1.232,
      "step": 291
    },
    {
      "epoch": 0.08512499088987684,
      "grad_norm": 23.749746678281905,
      "learning_rate": 9.921467638495666e-06,
      "loss": 1.0187,
      "step": 292
    },
    {
      "epoch": 0.08541651483128052,
      "grad_norm": 25.375333665812448,
      "learning_rate": 9.920632189069802e-06,
      "loss": 1.0857,
      "step": 293
    },
    {
      "epoch": 0.08570803877268421,
      "grad_norm": 24.15454126635134,
      "learning_rate": 9.919792354801614e-06,
      "loss": 0.9942,
      "step": 294
    },
    {
      "epoch": 0.08599956271408789,
      "grad_norm": 15.30287698911699,
      "learning_rate": 9.918948136439488e-06,
      "loss": 1.1961,
      "step": 295
    },
    {
      "epoch": 0.08629108665549158,
      "grad_norm": 58.17941828766543,
      "learning_rate": 9.91809953473572e-06,
      "loss": 1.2416,
      "step": 296
    },
    {
      "epoch": 0.08658261059689527,
      "grad_norm": 25.06678599355711,
      "learning_rate": 9.917246550446507e-06,
      "loss": 1.2373,
      "step": 297
    },
    {
      "epoch": 0.08687413453829895,
      "grad_norm": 26.620439422409564,
      "learning_rate": 9.916389184331957e-06,
      "loss": 1.149,
      "step": 298
    },
    {
      "epoch": 0.08716565847970265,
      "grad_norm": 24.75770831949848,
      "learning_rate": 9.915527437156083e-06,
      "loss": 1.1078,
      "step": 299
    },
    {
      "epoch": 0.08745718242110634,
      "grad_norm": 39.17661757917383,
      "learning_rate": 9.914661309686796e-06,
      "loss": 0.9873,
      "step": 300
    },
    {
      "epoch": 0.08745718242110634,
      "eval_loss": 1.209411859512329,
      "eval_runtime": 1960.3392,
      "eval_samples_per_second": 8.81,
      "eval_steps_per_second": 0.49,
      "step": 300
    },
    {
      "epoch": 0.08774870636251002,
      "grad_norm": 19.990656791020356,
      "learning_rate": 9.913790802695915e-06,
      "loss": 1.1954,
      "step": 301
    },
    {
      "epoch": 0.08804023030391371,
      "grad_norm": 24.87951805517374,
      "learning_rate": 9.912915916959162e-06,
      "loss": 1.1648,
      "step": 302
    },
    {
      "epoch": 0.08833175424531739,
      "grad_norm": 14.840381028356948,
      "learning_rate": 9.912036653256159e-06,
      "loss": 1.2417,
      "step": 303
    },
    {
      "epoch": 0.08862327818672108,
      "grad_norm": 18.45586120239632,
      "learning_rate": 9.911153012370427e-06,
      "loss": 1.1254,
      "step": 304
    },
    {
      "epoch": 0.08891480212812478,
      "grad_norm": 20.577776454892852,
      "learning_rate": 9.910264995089396e-06,
      "loss": 1.1993,
      "step": 305
    },
    {
      "epoch": 0.08920632606952845,
      "grad_norm": 15.336939861501197,
      "learning_rate": 9.909372602204385e-06,
      "loss": 1.156,
      "step": 306
    },
    {
      "epoch": 0.08949785001093215,
      "grad_norm": 22.5171733293677,
      "learning_rate": 9.90847583451062e-06,
      "loss": 1.054,
      "step": 307
    },
    {
      "epoch": 0.08978937395233584,
      "grad_norm": 20.326340867826637,
      "learning_rate": 9.907574692807223e-06,
      "loss": 1.0484,
      "step": 308
    },
    {
      "epoch": 0.09008089789373952,
      "grad_norm": 26.156308788019807,
      "learning_rate": 9.906669177897212e-06,
      "loss": 1.1078,
      "step": 309
    },
    {
      "epoch": 0.09037242183514321,
      "grad_norm": 79.99383185267934,
      "learning_rate": 9.905759290587506e-06,
      "loss": 1.0743,
      "step": 310
    },
    {
      "epoch": 0.09066394577654689,
      "grad_norm": 29.02414178680012,
      "learning_rate": 9.904845031688917e-06,
      "loss": 1.0648,
      "step": 311
    },
    {
      "epoch": 0.09095546971795058,
      "grad_norm": 69.51728109562505,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.9719,
      "step": 312
    },
    {
      "epoch": 0.09124699365935428,
      "grad_norm": 21.476229340680028,
      "learning_rate": 9.903003402387818e-06,
      "loss": 1.0727,
      "step": 313
    },
    {
      "epoch": 0.09153851760075796,
      "grad_norm": 20.51238914873807,
      "learning_rate": 9.902076033626409e-06,
      "loss": 1.1784,
      "step": 314
    },
    {
      "epoch": 0.09183004154216165,
      "grad_norm": 83.02307622653838,
      "learning_rate": 9.901144296558316e-06,
      "loss": 1.1091,
      "step": 315
    },
    {
      "epoch": 0.09212156548356534,
      "grad_norm": 29.141468496079387,
      "learning_rate": 9.900208192013825e-06,
      "loss": 1.1077,
      "step": 316
    },
    {
      "epoch": 0.09241308942496902,
      "grad_norm": 26.965109314780985,
      "learning_rate": 9.89926772082711e-06,
      "loss": 1.178,
      "step": 317
    },
    {
      "epoch": 0.09270461336637272,
      "grad_norm": 37.411535067471156,
      "learning_rate": 9.898322883836239e-06,
      "loss": 1.2114,
      "step": 318
    },
    {
      "epoch": 0.0929961373077764,
      "grad_norm": 95.9761434880239,
      "learning_rate": 9.897373681883167e-06,
      "loss": 1.0115,
      "step": 319
    },
    {
      "epoch": 0.09328766124918009,
      "grad_norm": 25.159532396091418,
      "learning_rate": 9.896420115813741e-06,
      "loss": 0.9719,
      "step": 320
    },
    {
      "epoch": 0.09357918519058378,
      "grad_norm": 26.403929689624267,
      "learning_rate": 9.8954621864777e-06,
      "loss": 1.3256,
      "step": 321
    },
    {
      "epoch": 0.09387070913198746,
      "grad_norm": 43.95358385627906,
      "learning_rate": 9.894499894728665e-06,
      "loss": 1.1627,
      "step": 322
    },
    {
      "epoch": 0.09416223307339115,
      "grad_norm": 59.270874822531084,
      "learning_rate": 9.893533241424144e-06,
      "loss": 1.2376,
      "step": 323
    },
    {
      "epoch": 0.09445375701479485,
      "grad_norm": 31.231099566436743,
      "learning_rate": 9.892562227425541e-06,
      "loss": 1.2669,
      "step": 324
    },
    {
      "epoch": 0.09474528095619852,
      "grad_norm": 30.000868290243098,
      "learning_rate": 9.891586853598139e-06,
      "loss": 0.8866,
      "step": 325
    },
    {
      "epoch": 0.09503680489760222,
      "grad_norm": 22.18767041089857,
      "learning_rate": 9.890607120811104e-06,
      "loss": 1.127,
      "step": 326
    },
    {
      "epoch": 0.09532832883900591,
      "grad_norm": 29.739003659877312,
      "learning_rate": 9.88962302993749e-06,
      "loss": 1.0286,
      "step": 327
    },
    {
      "epoch": 0.09561985278040959,
      "grad_norm": 54.81196664080981,
      "learning_rate": 9.888634581854235e-06,
      "loss": 1.1735,
      "step": 328
    },
    {
      "epoch": 0.09591137672181328,
      "grad_norm": 22.592341967073587,
      "learning_rate": 9.887641777442159e-06,
      "loss": 1.1049,
      "step": 329
    },
    {
      "epoch": 0.09620290066321696,
      "grad_norm": 17.326860340853685,
      "learning_rate": 9.88664461758596e-06,
      "loss": 0.885,
      "step": 330
    },
    {
      "epoch": 0.09649442460462065,
      "grad_norm": 26.74577155156345,
      "learning_rate": 9.885643103174224e-06,
      "loss": 1.1383,
      "step": 331
    },
    {
      "epoch": 0.09678594854602435,
      "grad_norm": 28.762893575583877,
      "learning_rate": 9.884637235099414e-06,
      "loss": 1.1292,
      "step": 332
    },
    {
      "epoch": 0.09707747248742803,
      "grad_norm": 16.207632465964565,
      "learning_rate": 9.88362701425787e-06,
      "loss": 1.0672,
      "step": 333
    },
    {
      "epoch": 0.09736899642883172,
      "grad_norm": 40.932229331768134,
      "learning_rate": 9.882612441549817e-06,
      "loss": 1.115,
      "step": 334
    },
    {
      "epoch": 0.09766052037023541,
      "grad_norm": 34.180380285967445,
      "learning_rate": 9.881593517879354e-06,
      "loss": 1.0895,
      "step": 335
    },
    {
      "epoch": 0.09795204431163909,
      "grad_norm": 26.188991740032947,
      "learning_rate": 9.880570244154455e-06,
      "loss": 1.1478,
      "step": 336
    },
    {
      "epoch": 0.09824356825304278,
      "grad_norm": 37.56875696108895,
      "learning_rate": 9.879542621286975e-06,
      "loss": 1.1567,
      "step": 337
    },
    {
      "epoch": 0.09853509219444646,
      "grad_norm": 38.74715923165894,
      "learning_rate": 9.878510650192644e-06,
      "loss": 1.2177,
      "step": 338
    },
    {
      "epoch": 0.09882661613585016,
      "grad_norm": 30.415450440325703,
      "learning_rate": 9.877474331791063e-06,
      "loss": 1.1205,
      "step": 339
    },
    {
      "epoch": 0.09911814007725385,
      "grad_norm": 29.38628858565479,
      "learning_rate": 9.876433667005711e-06,
      "loss": 1.0952,
      "step": 340
    },
    {
      "epoch": 0.09940966401865753,
      "grad_norm": 28.57147802028096,
      "learning_rate": 9.875388656763937e-06,
      "loss": 1.2016,
      "step": 341
    },
    {
      "epoch": 0.09970118796006122,
      "grad_norm": 31.093350027885965,
      "learning_rate": 9.874339301996968e-06,
      "loss": 1.2165,
      "step": 342
    },
    {
      "epoch": 0.09999271190146491,
      "grad_norm": 23.350944807826,
      "learning_rate": 9.873285603639893e-06,
      "loss": 1.0348,
      "step": 343
    },
    {
      "epoch": 0.1002842358428686,
      "grad_norm": 37.66943334568452,
      "learning_rate": 9.87222756263168e-06,
      "loss": 1.0425,
      "step": 344
    },
    {
      "epoch": 0.10057575978427229,
      "grad_norm": 59.31839716197531,
      "learning_rate": 9.871165179915162e-06,
      "loss": 1.2543,
      "step": 345
    },
    {
      "epoch": 0.10086728372567597,
      "grad_norm": 45.729186453996675,
      "learning_rate": 9.870098456437045e-06,
      "loss": 1.2108,
      "step": 346
    },
    {
      "epoch": 0.10115880766707966,
      "grad_norm": 21.3106886880423,
      "learning_rate": 9.869027393147898e-06,
      "loss": 1.2037,
      "step": 347
    },
    {
      "epoch": 0.10145033160848335,
      "grad_norm": 33.35909803941295,
      "learning_rate": 9.867951991002162e-06,
      "loss": 1.0783,
      "step": 348
    },
    {
      "epoch": 0.10174185554988703,
      "grad_norm": 91.77423957002037,
      "learning_rate": 9.866872250958144e-06,
      "loss": 0.9938,
      "step": 349
    },
    {
      "epoch": 0.10203337949129072,
      "grad_norm": 183.74345424169522,
      "learning_rate": 9.865788173978011e-06,
      "loss": 1.0394,
      "step": 350
    },
    {
      "epoch": 0.10203337949129072,
      "eval_loss": 1.1943471431732178,
      "eval_runtime": 1940.1514,
      "eval_samples_per_second": 8.901,
      "eval_steps_per_second": 0.495,
      "step": 350
    },
    {
      "epoch": 0.10232490343269442,
      "grad_norm": 55.586957566979066,
      "learning_rate": 9.864699761027801e-06,
      "loss": 1.068,
      "step": 351
    },
    {
      "epoch": 0.1026164273740981,
      "grad_norm": 24.93511476301786,
      "learning_rate": 9.863607013077414e-06,
      "loss": 1.0642,
      "step": 352
    },
    {
      "epoch": 0.10290795131550179,
      "grad_norm": 26.798308032456326,
      "learning_rate": 9.862509931100612e-06,
      "loss": 1.2114,
      "step": 353
    },
    {
      "epoch": 0.10319947525690547,
      "grad_norm": 88.61385747340748,
      "learning_rate": 9.86140851607502e-06,
      "loss": 1.044,
      "step": 354
    },
    {
      "epoch": 0.10349099919830916,
      "grad_norm": 18.65680105540008,
      "learning_rate": 9.860302768982124e-06,
      "loss": 0.9915,
      "step": 355
    },
    {
      "epoch": 0.10378252313971285,
      "grad_norm": 24.181335829527125,
      "learning_rate": 9.85919269080727e-06,
      "loss": 1.0264,
      "step": 356
    },
    {
      "epoch": 0.10407404708111653,
      "grad_norm": 57.743308899487936,
      "learning_rate": 9.858078282539662e-06,
      "loss": 0.9493,
      "step": 357
    },
    {
      "epoch": 0.10436557102252023,
      "grad_norm": 55.02387861486096,
      "learning_rate": 9.856959545172369e-06,
      "loss": 1.2679,
      "step": 358
    },
    {
      "epoch": 0.10465709496392392,
      "grad_norm": 84.9839410603266,
      "learning_rate": 9.855836479702307e-06,
      "loss": 1.1779,
      "step": 359
    },
    {
      "epoch": 0.1049486189053276,
      "grad_norm": 18.522845500748478,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.9399,
      "step": 360
    },
    {
      "epoch": 0.10524014284673129,
      "grad_norm": 22.063517567389727,
      "learning_rate": 9.853577368460862e-06,
      "loss": 0.9988,
      "step": 361
    },
    {
      "epoch": 0.10553166678813497,
      "grad_norm": 34.99786943590083,
      "learning_rate": 9.852441324702599e-06,
      "loss": 1.2182,
      "step": 362
    },
    {
      "epoch": 0.10582319072953866,
      "grad_norm": 78.8545075988515,
      "learning_rate": 9.851300956867821e-06,
      "loss": 1.1577,
      "step": 363
    },
    {
      "epoch": 0.10611471467094236,
      "grad_norm": 32.30348455176298,
      "learning_rate": 9.850156265972722e-06,
      "loss": 1.1389,
      "step": 364
    },
    {
      "epoch": 0.10640623861234604,
      "grad_norm": 45.954820690599206,
      "learning_rate": 9.84900725303735e-06,
      "loss": 1.2109,
      "step": 365
    },
    {
      "epoch": 0.10669776255374973,
      "grad_norm": 20.239071812465276,
      "learning_rate": 9.847853919085608e-06,
      "loss": 0.9637,
      "step": 366
    },
    {
      "epoch": 0.10698928649515342,
      "grad_norm": 24.61713108243007,
      "learning_rate": 9.846696265145249e-06,
      "loss": 1.1903,
      "step": 367
    },
    {
      "epoch": 0.1072808104365571,
      "grad_norm": 33.224395504948696,
      "learning_rate": 9.845534292247872e-06,
      "loss": 1.1423,
      "step": 368
    },
    {
      "epoch": 0.1075723343779608,
      "grad_norm": 56.75773173782299,
      "learning_rate": 9.844368001428929e-06,
      "loss": 1.1787,
      "step": 369
    },
    {
      "epoch": 0.10786385831936447,
      "grad_norm": 35.15628750378441,
      "learning_rate": 9.843197393727713e-06,
      "loss": 0.9867,
      "step": 370
    },
    {
      "epoch": 0.10815538226076817,
      "grad_norm": 78.2993118996902,
      "learning_rate": 9.842022470187374e-06,
      "loss": 1.181,
      "step": 371
    },
    {
      "epoch": 0.10844690620217186,
      "grad_norm": 54.106978658256295,
      "learning_rate": 9.8408432318549e-06,
      "loss": 1.2124,
      "step": 372
    },
    {
      "epoch": 0.10873843014357554,
      "grad_norm": 30.892063825736635,
      "learning_rate": 9.839659679781126e-06,
      "loss": 1.2256,
      "step": 373
    },
    {
      "epoch": 0.10902995408497923,
      "grad_norm": 31.67964626337866,
      "learning_rate": 9.838471815020731e-06,
      "loss": 1.1513,
      "step": 374
    },
    {
      "epoch": 0.10932147802638292,
      "grad_norm": 71.2934874022383,
      "learning_rate": 9.83727963863224e-06,
      "loss": 1.1654,
      "step": 375
    },
    {
      "epoch": 0.1096130019677866,
      "grad_norm": 29.7136552558913,
      "learning_rate": 9.836083151678014e-06,
      "loss": 1.0958,
      "step": 376
    },
    {
      "epoch": 0.1099045259091903,
      "grad_norm": 61.63525983126351,
      "learning_rate": 9.834882355224261e-06,
      "loss": 0.9496,
      "step": 377
    },
    {
      "epoch": 0.11019604985059397,
      "grad_norm": 27.648432599478166,
      "learning_rate": 9.833677250341027e-06,
      "loss": 1.0862,
      "step": 378
    },
    {
      "epoch": 0.11048757379199767,
      "grad_norm": 26.583718893564217,
      "learning_rate": 9.832467838102197e-06,
      "loss": 1.2242,
      "step": 379
    },
    {
      "epoch": 0.11077909773340136,
      "grad_norm": 18.738157474334784,
      "learning_rate": 9.831254119585497e-06,
      "loss": 1.0358,
      "step": 380
    },
    {
      "epoch": 0.11107062167480504,
      "grad_norm": 58.65725327371997,
      "learning_rate": 9.830036095872482e-06,
      "loss": 1.1383,
      "step": 381
    },
    {
      "epoch": 0.11136214561620873,
      "grad_norm": 22.52130431539307,
      "learning_rate": 9.828813768048555e-06,
      "loss": 1.1693,
      "step": 382
    },
    {
      "epoch": 0.11165366955761243,
      "grad_norm": 22.791622703684848,
      "learning_rate": 9.827587137202947e-06,
      "loss": 1.136,
      "step": 383
    },
    {
      "epoch": 0.1119451934990161,
      "grad_norm": 14.591566245088888,
      "learning_rate": 9.826356204428726e-06,
      "loss": 0.9395,
      "step": 384
    },
    {
      "epoch": 0.1122367174404198,
      "grad_norm": 37.93271079949477,
      "learning_rate": 9.825120970822794e-06,
      "loss": 1.0666,
      "step": 385
    },
    {
      "epoch": 0.11252824138182348,
      "grad_norm": 45.821742876845725,
      "learning_rate": 9.823881437485882e-06,
      "loss": 0.9187,
      "step": 386
    },
    {
      "epoch": 0.11281976532322717,
      "grad_norm": 25.048313940190774,
      "learning_rate": 9.822637605522556e-06,
      "loss": 1.0076,
      "step": 387
    },
    {
      "epoch": 0.11311128926463086,
      "grad_norm": 21.975725391969096,
      "learning_rate": 9.821389476041212e-06,
      "loss": 0.9247,
      "step": 388
    },
    {
      "epoch": 0.11340281320603454,
      "grad_norm": 21.92328307236672,
      "learning_rate": 9.820137050154077e-06,
      "loss": 1.0046,
      "step": 389
    },
    {
      "epoch": 0.11369433714743823,
      "grad_norm": 61.148858612220465,
      "learning_rate": 9.8188803289772e-06,
      "loss": 1.0963,
      "step": 390
    },
    {
      "epoch": 0.11398586108884193,
      "grad_norm": 20.05038478065898,
      "learning_rate": 9.817619313630467e-06,
      "loss": 1.0079,
      "step": 391
    },
    {
      "epoch": 0.1142773850302456,
      "grad_norm": 18.53862316507373,
      "learning_rate": 9.816354005237583e-06,
      "loss": 1.2539,
      "step": 392
    },
    {
      "epoch": 0.1145689089716493,
      "grad_norm": 34.48641432543646,
      "learning_rate": 9.815084404926082e-06,
      "loss": 0.9887,
      "step": 393
    },
    {
      "epoch": 0.11486043291305298,
      "grad_norm": 25.981936180223133,
      "learning_rate": 9.813810513827324e-06,
      "loss": 1.2207,
      "step": 394
    },
    {
      "epoch": 0.11515195685445667,
      "grad_norm": 22.563655537150797,
      "learning_rate": 9.812532333076488e-06,
      "loss": 1.0975,
      "step": 395
    },
    {
      "epoch": 0.11544348079586036,
      "grad_norm": 23.710868363329034,
      "learning_rate": 9.811249863812581e-06,
      "loss": 1.15,
      "step": 396
    },
    {
      "epoch": 0.11573500473726404,
      "grad_norm": 39.367027861515396,
      "learning_rate": 9.809963107178426e-06,
      "loss": 1.1004,
      "step": 397
    },
    {
      "epoch": 0.11602652867866774,
      "grad_norm": 36.53853978727232,
      "learning_rate": 9.808672064320672e-06,
      "loss": 0.9546,
      "step": 398
    },
    {
      "epoch": 0.11631805262007143,
      "grad_norm": 23.412613415050036,
      "learning_rate": 9.807376736389785e-06,
      "loss": 1.2499,
      "step": 399
    },
    {
      "epoch": 0.11660957656147511,
      "grad_norm": 27.168461924810845,
      "learning_rate": 9.806077124540045e-06,
      "loss": 1.0795,
      "step": 400
    },
    {
      "epoch": 0.11660957656147511,
      "eval_loss": 1.1913939714431763,
      "eval_runtime": 1933.6356,
      "eval_samples_per_second": 8.931,
      "eval_steps_per_second": 0.496,
      "step": 400
    },
    {
      "epoch": 0.1169011005028788,
      "grad_norm": 33.67781948124961,
      "learning_rate": 9.804773229929558e-06,
      "loss": 0.9511,
      "step": 401
    },
    {
      "epoch": 0.11719262444428248,
      "grad_norm": 31.748294242324427,
      "learning_rate": 9.803465053720242e-06,
      "loss": 0.9602,
      "step": 402
    },
    {
      "epoch": 0.11748414838568617,
      "grad_norm": 54.392212649908586,
      "learning_rate": 9.80215259707783e-06,
      "loss": 1.2041,
      "step": 403
    },
    {
      "epoch": 0.11777567232708987,
      "grad_norm": 48.12736490412038,
      "learning_rate": 9.800835861171869e-06,
      "loss": 1.2717,
      "step": 404
    },
    {
      "epoch": 0.11806719626849355,
      "grad_norm": 17.391337050492155,
      "learning_rate": 9.799514847175721e-06,
      "loss": 1.1562,
      "step": 405
    },
    {
      "epoch": 0.11835872020989724,
      "grad_norm": 44.91068216915621,
      "learning_rate": 9.798189556266559e-06,
      "loss": 1.2378,
      "step": 406
    },
    {
      "epoch": 0.11865024415130093,
      "grad_norm": 30.054656881776285,
      "learning_rate": 9.79685998962537e-06,
      "loss": 1.2251,
      "step": 407
    },
    {
      "epoch": 0.11894176809270461,
      "grad_norm": 29.87298217413138,
      "learning_rate": 9.795526148436945e-06,
      "loss": 1.0173,
      "step": 408
    },
    {
      "epoch": 0.1192332920341083,
      "grad_norm": 52.44711107967395,
      "learning_rate": 9.79418803388989e-06,
      "loss": 1.1766,
      "step": 409
    },
    {
      "epoch": 0.11952481597551198,
      "grad_norm": 30.99739492350727,
      "learning_rate": 9.792845647176621e-06,
      "loss": 1.0345,
      "step": 410
    },
    {
      "epoch": 0.11981633991691568,
      "grad_norm": 64.39630225652625,
      "learning_rate": 9.79149898949335e-06,
      "loss": 1.1109,
      "step": 411
    },
    {
      "epoch": 0.12010786385831937,
      "grad_norm": 17.472005901965126,
      "learning_rate": 9.790148062040108e-06,
      "loss": 1.0195,
      "step": 412
    },
    {
      "epoch": 0.12039938779972305,
      "grad_norm": 44.925334867163585,
      "learning_rate": 9.78879286602072e-06,
      "loss": 1.3001,
      "step": 413
    },
    {
      "epoch": 0.12069091174112674,
      "grad_norm": 26.20291485611835,
      "learning_rate": 9.787433402642823e-06,
      "loss": 1.0204,
      "step": 414
    },
    {
      "epoch": 0.12098243568253043,
      "grad_norm": 47.37675721945227,
      "learning_rate": 9.786069673117852e-06,
      "loss": 1.0852,
      "step": 415
    },
    {
      "epoch": 0.12127395962393411,
      "grad_norm": 40.02496704948362,
      "learning_rate": 9.784701678661045e-06,
      "loss": 1.068,
      "step": 416
    },
    {
      "epoch": 0.1215654835653378,
      "grad_norm": 28.327806461475983,
      "learning_rate": 9.78332942049144e-06,
      "loss": 1.1381,
      "step": 417
    },
    {
      "epoch": 0.12185700750674149,
      "grad_norm": 51.081740609570744,
      "learning_rate": 9.781952899831876e-06,
      "loss": 1.1033,
      "step": 418
    },
    {
      "epoch": 0.12214853144814518,
      "grad_norm": 20.138531716186357,
      "learning_rate": 9.780572117908991e-06,
      "loss": 1.1349,
      "step": 419
    },
    {
      "epoch": 0.12244005538954887,
      "grad_norm": 27.47765711377529,
      "learning_rate": 9.779187075953215e-06,
      "loss": 1.1085,
      "step": 420
    },
    {
      "epoch": 0.12273157933095255,
      "grad_norm": 25.585382604715623,
      "learning_rate": 9.777797775198778e-06,
      "loss": 1.0028,
      "step": 421
    },
    {
      "epoch": 0.12302310327235624,
      "grad_norm": 34.13007717597658,
      "learning_rate": 9.776404216883709e-06,
      "loss": 1.0205,
      "step": 422
    },
    {
      "epoch": 0.12331462721375994,
      "grad_norm": 13.913248203075367,
      "learning_rate": 9.775006402249821e-06,
      "loss": 0.9342,
      "step": 423
    },
    {
      "epoch": 0.12360615115516362,
      "grad_norm": 26.72790274976063,
      "learning_rate": 9.77360433254273e-06,
      "loss": 1.0229,
      "step": 424
    },
    {
      "epoch": 0.12389767509656731,
      "grad_norm": 20.682382553758192,
      "learning_rate": 9.772198009011837e-06,
      "loss": 0.9803,
      "step": 425
    },
    {
      "epoch": 0.12418919903797099,
      "grad_norm": 29.88386857737213,
      "learning_rate": 9.770787432910336e-06,
      "loss": 0.9566,
      "step": 426
    },
    {
      "epoch": 0.12448072297937468,
      "grad_norm": 18.875233586271055,
      "learning_rate": 9.769372605495212e-06,
      "loss": 1.1362,
      "step": 427
    },
    {
      "epoch": 0.12477224692077837,
      "grad_norm": 34.235877647879036,
      "learning_rate": 9.767953528027238e-06,
      "loss": 0.9484,
      "step": 428
    },
    {
      "epoch": 0.12506377086218207,
      "grad_norm": 24.96792967621483,
      "learning_rate": 9.766530201770969e-06,
      "loss": 1.0221,
      "step": 429
    },
    {
      "epoch": 0.12535529480358573,
      "grad_norm": 32.33288198672142,
      "learning_rate": 9.765102627994757e-06,
      "loss": 1.2214,
      "step": 430
    },
    {
      "epoch": 0.12564681874498942,
      "grad_norm": 23.225117773273322,
      "learning_rate": 9.763670807970724e-06,
      "loss": 1.0578,
      "step": 431
    },
    {
      "epoch": 0.12593834268639312,
      "grad_norm": 33.37162123603574,
      "learning_rate": 9.762234742974793e-06,
      "loss": 1.0154,
      "step": 432
    },
    {
      "epoch": 0.1262298666277968,
      "grad_norm": 68.3375859656722,
      "learning_rate": 9.760794434286653e-06,
      "loss": 1.0641,
      "step": 433
    },
    {
      "epoch": 0.1265213905692005,
      "grad_norm": 51.49147394333224,
      "learning_rate": 9.759349883189788e-06,
      "loss": 1.0153,
      "step": 434
    },
    {
      "epoch": 0.1268129145106042,
      "grad_norm": 38.81290717760302,
      "learning_rate": 9.757901090971458e-06,
      "loss": 0.926,
      "step": 435
    },
    {
      "epoch": 0.12710443845200786,
      "grad_norm": 27.12351497815414,
      "learning_rate": 9.756448058922697e-06,
      "loss": 1.0985,
      "step": 436
    },
    {
      "epoch": 0.12739596239341155,
      "grad_norm": 20.54629073273183,
      "learning_rate": 9.754990788338323e-06,
      "loss": 1.1426,
      "step": 437
    },
    {
      "epoch": 0.12768748633481525,
      "grad_norm": 19.575916802795373,
      "learning_rate": 9.753529280516931e-06,
      "loss": 1.0602,
      "step": 438
    },
    {
      "epoch": 0.12797901027621894,
      "grad_norm": 24.92411768911073,
      "learning_rate": 9.752063536760892e-06,
      "loss": 1.0978,
      "step": 439
    },
    {
      "epoch": 0.12827053421762263,
      "grad_norm": 55.00377486652154,
      "learning_rate": 9.750593558376347e-06,
      "loss": 1.1176,
      "step": 440
    },
    {
      "epoch": 0.1285620581590263,
      "grad_norm": 41.80306139904889,
      "learning_rate": 9.749119346673214e-06,
      "loss": 1.0692,
      "step": 441
    },
    {
      "epoch": 0.12885358210043,
      "grad_norm": 32.43062837101524,
      "learning_rate": 9.747640902965185e-06,
      "loss": 1.3012,
      "step": 442
    },
    {
      "epoch": 0.12914510604183368,
      "grad_norm": 16.76351721145627,
      "learning_rate": 9.74615822856972e-06,
      "loss": 1.0007,
      "step": 443
    },
    {
      "epoch": 0.12943662998323738,
      "grad_norm": 24.148937733165017,
      "learning_rate": 9.74467132480805e-06,
      "loss": 1.0631,
      "step": 444
    },
    {
      "epoch": 0.12972815392464107,
      "grad_norm": 22.37403997827686,
      "learning_rate": 9.743180193005177e-06,
      "loss": 1.1331,
      "step": 445
    },
    {
      "epoch": 0.13001967786604474,
      "grad_norm": 22.260663932531614,
      "learning_rate": 9.741684834489866e-06,
      "loss": 1.1259,
      "step": 446
    },
    {
      "epoch": 0.13031120180744843,
      "grad_norm": 84.83255423231579,
      "learning_rate": 9.740185250594656e-06,
      "loss": 0.995,
      "step": 447
    },
    {
      "epoch": 0.13060272574885212,
      "grad_norm": 29.488435398882505,
      "learning_rate": 9.738681442655842e-06,
      "loss": 1.0828,
      "step": 448
    },
    {
      "epoch": 0.13089424969025582,
      "grad_norm": 50.80562905761197,
      "learning_rate": 9.737173412013488e-06,
      "loss": 1.1732,
      "step": 449
    },
    {
      "epoch": 0.1311857736316595,
      "grad_norm": 23.155488033668984,
      "learning_rate": 9.735661160011424e-06,
      "loss": 1.0437,
      "step": 450
    },
    {
      "epoch": 0.1311857736316595,
      "eval_loss": 1.1842231750488281,
      "eval_runtime": 1960.7873,
      "eval_samples_per_second": 8.808,
      "eval_steps_per_second": 0.49,
      "step": 450
    },
    {
      "epoch": 0.1314772975730632,
      "grad_norm": 52.618212841333396,
      "learning_rate": 9.734144687997235e-06,
      "loss": 1.1746,
      "step": 451
    },
    {
      "epoch": 0.13176882151446687,
      "grad_norm": 32.17090408262094,
      "learning_rate": 9.732623997322274e-06,
      "loss": 1.1758,
      "step": 452
    },
    {
      "epoch": 0.13206034545587056,
      "grad_norm": 36.90946535083571,
      "learning_rate": 9.731099089341645e-06,
      "loss": 1.0738,
      "step": 453
    },
    {
      "epoch": 0.13235186939727425,
      "grad_norm": 24.542376264916403,
      "learning_rate": 9.729569965414214e-06,
      "loss": 1.1829,
      "step": 454
    },
    {
      "epoch": 0.13264339333867795,
      "grad_norm": 22.63506255068065,
      "learning_rate": 9.728036626902607e-06,
      "loss": 1.0462,
      "step": 455
    },
    {
      "epoch": 0.13293491728008164,
      "grad_norm": 28.523287776068603,
      "learning_rate": 9.726499075173201e-06,
      "loss": 1.1686,
      "step": 456
    },
    {
      "epoch": 0.1332264412214853,
      "grad_norm": 33.3801925779369,
      "learning_rate": 9.724957311596132e-06,
      "loss": 1.0695,
      "step": 457
    },
    {
      "epoch": 0.133517965162889,
      "grad_norm": 40.89915588689837,
      "learning_rate": 9.723411337545283e-06,
      "loss": 1.112,
      "step": 458
    },
    {
      "epoch": 0.1338094891042927,
      "grad_norm": 39.62733674522131,
      "learning_rate": 9.721861154398293e-06,
      "loss": 1.0333,
      "step": 459
    },
    {
      "epoch": 0.13410101304569638,
      "grad_norm": 21.358773488176357,
      "learning_rate": 9.720306763536553e-06,
      "loss": 1.2569,
      "step": 460
    },
    {
      "epoch": 0.13439253698710008,
      "grad_norm": 68.59005641622537,
      "learning_rate": 9.718748166345203e-06,
      "loss": 1.0723,
      "step": 461
    },
    {
      "epoch": 0.13468406092850374,
      "grad_norm": 54.21765217860693,
      "learning_rate": 9.717185364213127e-06,
      "loss": 1.0916,
      "step": 462
    },
    {
      "epoch": 0.13497558486990743,
      "grad_norm": 98.83330105151403,
      "learning_rate": 9.71561835853296e-06,
      "loss": 0.9647,
      "step": 463
    },
    {
      "epoch": 0.13526710881131113,
      "grad_norm": 23.781282568981805,
      "learning_rate": 9.714047150701082e-06,
      "loss": 1.1311,
      "step": 464
    },
    {
      "epoch": 0.13555863275271482,
      "grad_norm": 33.82488104178976,
      "learning_rate": 9.712471742117618e-06,
      "loss": 1.1552,
      "step": 465
    },
    {
      "epoch": 0.1358501566941185,
      "grad_norm": 103.98485675952259,
      "learning_rate": 9.710892134186438e-06,
      "loss": 1.2213,
      "step": 466
    },
    {
      "epoch": 0.1361416806355222,
      "grad_norm": 68.85958780917224,
      "learning_rate": 9.70930832831515e-06,
      "loss": 1.0322,
      "step": 467
    },
    {
      "epoch": 0.13643320457692587,
      "grad_norm": 36.484320933614114,
      "learning_rate": 9.707720325915105e-06,
      "loss": 1.0214,
      "step": 468
    },
    {
      "epoch": 0.13672472851832956,
      "grad_norm": 22.88191388100698,
      "learning_rate": 9.706128128401394e-06,
      "loss": 0.9163,
      "step": 469
    },
    {
      "epoch": 0.13701625245973326,
      "grad_norm": 19.3569483589172,
      "learning_rate": 9.704531737192847e-06,
      "loss": 1.0084,
      "step": 470
    },
    {
      "epoch": 0.13730777640113695,
      "grad_norm": 20.139816202499254,
      "learning_rate": 9.702931153712027e-06,
      "loss": 0.9683,
      "step": 471
    },
    {
      "epoch": 0.13759930034254064,
      "grad_norm": 73.43580352802837,
      "learning_rate": 9.701326379385238e-06,
      "loss": 1.147,
      "step": 472
    },
    {
      "epoch": 0.1378908242839443,
      "grad_norm": 15.205496121514392,
      "learning_rate": 9.699717415642518e-06,
      "loss": 1.0702,
      "step": 473
    },
    {
      "epoch": 0.138182348225348,
      "grad_norm": 32.218853243111674,
      "learning_rate": 9.698104263917632e-06,
      "loss": 1.1657,
      "step": 474
    },
    {
      "epoch": 0.1384738721667517,
      "grad_norm": 17.657363537058487,
      "learning_rate": 9.696486925648084e-06,
      "loss": 1.176,
      "step": 475
    },
    {
      "epoch": 0.1387653961081554,
      "grad_norm": 29.732437032712248,
      "learning_rate": 9.694865402275105e-06,
      "loss": 1.0969,
      "step": 476
    },
    {
      "epoch": 0.13905692004955908,
      "grad_norm": 21.621500876448017,
      "learning_rate": 9.69323969524366e-06,
      "loss": 0.989,
      "step": 477
    },
    {
      "epoch": 0.13934844399096274,
      "grad_norm": 17.941755451659716,
      "learning_rate": 9.691609806002433e-06,
      "loss": 1.0331,
      "step": 478
    },
    {
      "epoch": 0.13963996793236644,
      "grad_norm": 35.63608229584576,
      "learning_rate": 9.689975736003845e-06,
      "loss": 1.189,
      "step": 479
    },
    {
      "epoch": 0.13993149187377013,
      "grad_norm": 29.40381590077013,
      "learning_rate": 9.688337486704038e-06,
      "loss": 1.0782,
      "step": 480
    },
    {
      "epoch": 0.14022301581517382,
      "grad_norm": 44.9733290090396,
      "learning_rate": 9.686695059562875e-06,
      "loss": 1.1944,
      "step": 481
    },
    {
      "epoch": 0.14051453975657752,
      "grad_norm": 39.56314814055381,
      "learning_rate": 9.68504845604395e-06,
      "loss": 1.16,
      "step": 482
    },
    {
      "epoch": 0.1408060636979812,
      "grad_norm": 58.60760885711245,
      "learning_rate": 9.683397677614567e-06,
      "loss": 0.9023,
      "step": 483
    },
    {
      "epoch": 0.14109758763938487,
      "grad_norm": 88.54692894813985,
      "learning_rate": 9.681742725745762e-06,
      "loss": 0.9968,
      "step": 484
    },
    {
      "epoch": 0.14138911158078857,
      "grad_norm": 33.50404206552957,
      "learning_rate": 9.680083601912284e-06,
      "loss": 1.1379,
      "step": 485
    },
    {
      "epoch": 0.14168063552219226,
      "grad_norm": 42.5287051440329,
      "learning_rate": 9.678420307592602e-06,
      "loss": 1.1475,
      "step": 486
    },
    {
      "epoch": 0.14197215946359595,
      "grad_norm": 32.60532786776747,
      "learning_rate": 9.6767528442689e-06,
      "loss": 1.0916,
      "step": 487
    },
    {
      "epoch": 0.14226368340499965,
      "grad_norm": 18.93522671835132,
      "learning_rate": 9.675081213427076e-06,
      "loss": 1.2279,
      "step": 488
    },
    {
      "epoch": 0.1425552073464033,
      "grad_norm": 22.616518588742462,
      "learning_rate": 9.673405416556741e-06,
      "loss": 1.1132,
      "step": 489
    },
    {
      "epoch": 0.142846731287807,
      "grad_norm": 27.032003681784907,
      "learning_rate": 9.671725455151226e-06,
      "loss": 1.0691,
      "step": 490
    },
    {
      "epoch": 0.1431382552292107,
      "grad_norm": 40.67878550748602,
      "learning_rate": 9.670041330707563e-06,
      "loss": 1.0076,
      "step": 491
    },
    {
      "epoch": 0.1434297791706144,
      "grad_norm": 47.468173833057655,
      "learning_rate": 9.668353044726498e-06,
      "loss": 1.2926,
      "step": 492
    },
    {
      "epoch": 0.14372130311201808,
      "grad_norm": 36.01871208765648,
      "learning_rate": 9.66666059871249e-06,
      "loss": 1.1484,
      "step": 493
    },
    {
      "epoch": 0.14401282705342175,
      "grad_norm": 19.593194701286706,
      "learning_rate": 9.664963994173695e-06,
      "loss": 1.0079,
      "step": 494
    },
    {
      "epoch": 0.14430435099482544,
      "grad_norm": 82.0795980159821,
      "learning_rate": 9.663263232621985e-06,
      "loss": 0.9896,
      "step": 495
    },
    {
      "epoch": 0.14459587493622914,
      "grad_norm": 41.609418502308166,
      "learning_rate": 9.66155831557293e-06,
      "loss": 0.9735,
      "step": 496
    },
    {
      "epoch": 0.14488739887763283,
      "grad_norm": 39.02412397832953,
      "learning_rate": 9.659849244545804e-06,
      "loss": 0.8294,
      "step": 497
    },
    {
      "epoch": 0.14517892281903652,
      "grad_norm": 22.256682846774307,
      "learning_rate": 9.658136021063585e-06,
      "loss": 0.8812,
      "step": 498
    },
    {
      "epoch": 0.14547044676044021,
      "grad_norm": 28.810278134436665,
      "learning_rate": 9.656418646652947e-06,
      "loss": 1.1473,
      "step": 499
    },
    {
      "epoch": 0.14576197070184388,
      "grad_norm": 24.122733613502234,
      "learning_rate": 9.65469712284427e-06,
      "loss": 1.1184,
      "step": 500
    },
    {
      "epoch": 0.14576197070184388,
      "eval_loss": 1.180803656578064,
      "eval_runtime": 1981.6204,
      "eval_samples_per_second": 8.715,
      "eval_steps_per_second": 0.484,
      "step": 500
    },
    {
      "epoch": 0.14605349464324757,
      "grad_norm": 38.2884266955853,
      "learning_rate": 9.652971451171625e-06,
      "loss": 1.2243,
      "step": 501
    },
    {
      "epoch": 0.14634501858465127,
      "grad_norm": 35.43952404347885,
      "learning_rate": 9.651241633172782e-06,
      "loss": 0.9481,
      "step": 502
    },
    {
      "epoch": 0.14663654252605496,
      "grad_norm": 82.55353108504433,
      "learning_rate": 9.649507670389202e-06,
      "loss": 0.9949,
      "step": 503
    },
    {
      "epoch": 0.14692806646745865,
      "grad_norm": 59.10796729887697,
      "learning_rate": 9.647769564366048e-06,
      "loss": 1.0925,
      "step": 504
    },
    {
      "epoch": 0.14721959040886232,
      "grad_norm": 19.800404591484593,
      "learning_rate": 9.646027316652167e-06,
      "loss": 1.017,
      "step": 505
    },
    {
      "epoch": 0.147511114350266,
      "grad_norm": 43.00042930063944,
      "learning_rate": 9.644280928800101e-06,
      "loss": 1.0631,
      "step": 506
    },
    {
      "epoch": 0.1478026382916697,
      "grad_norm": 26.955303110572224,
      "learning_rate": 9.64253040236608e-06,
      "loss": 1.0236,
      "step": 507
    },
    {
      "epoch": 0.1480941622330734,
      "grad_norm": 31.841839642393623,
      "learning_rate": 9.640775738910019e-06,
      "loss": 1.1341,
      "step": 508
    },
    {
      "epoch": 0.1483856861744771,
      "grad_norm": 29.074873437298397,
      "learning_rate": 9.639016939995526e-06,
      "loss": 0.9046,
      "step": 509
    },
    {
      "epoch": 0.14867721011588075,
      "grad_norm": 32.01930500239041,
      "learning_rate": 9.63725400718989e-06,
      "loss": 0.9523,
      "step": 510
    },
    {
      "epoch": 0.14896873405728445,
      "grad_norm": 37.044162929391526,
      "learning_rate": 9.635486942064082e-06,
      "loss": 1.0895,
      "step": 511
    },
    {
      "epoch": 0.14926025799868814,
      "grad_norm": 21.17557925948679,
      "learning_rate": 9.633715746192762e-06,
      "loss": 1.0886,
      "step": 512
    },
    {
      "epoch": 0.14955178194009183,
      "grad_norm": 22.669418646660557,
      "learning_rate": 9.631940421154263e-06,
      "loss": 1.1533,
      "step": 513
    },
    {
      "epoch": 0.14984330588149553,
      "grad_norm": 21.651318804068516,
      "learning_rate": 9.630160968530601e-06,
      "loss": 0.8781,
      "step": 514
    },
    {
      "epoch": 0.15013482982289922,
      "grad_norm": 70.61270504863928,
      "learning_rate": 9.628377389907475e-06,
      "loss": 1.1256,
      "step": 515
    },
    {
      "epoch": 0.15042635376430288,
      "grad_norm": 28.116434879905697,
      "learning_rate": 9.626589686874252e-06,
      "loss": 1.0099,
      "step": 516
    },
    {
      "epoch": 0.15071787770570658,
      "grad_norm": 28.97689440759099,
      "learning_rate": 9.62479786102398e-06,
      "loss": 1.1795,
      "step": 517
    },
    {
      "epoch": 0.15100940164711027,
      "grad_norm": 25.271330555530895,
      "learning_rate": 9.62300191395338e-06,
      "loss": 0.961,
      "step": 518
    },
    {
      "epoch": 0.15130092558851396,
      "grad_norm": 29.636348152693085,
      "learning_rate": 9.621201847262844e-06,
      "loss": 1.1018,
      "step": 519
    },
    {
      "epoch": 0.15159244952991766,
      "grad_norm": 30.574751206610358,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.8872,
      "step": 520
    },
    {
      "epoch": 0.15188397347132132,
      "grad_norm": 23.598450529809195,
      "learning_rate": 9.617589361441889e-06,
      "loss": 0.9842,
      "step": 521
    },
    {
      "epoch": 0.152175497412725,
      "grad_norm": 54.77930724755095,
      "learning_rate": 9.615776945530603e-06,
      "loss": 1.0471,
      "step": 522
    },
    {
      "epoch": 0.1524670213541287,
      "grad_norm": 19.640809507135764,
      "learning_rate": 9.613960416437652e-06,
      "loss": 1.1437,
      "step": 523
    },
    {
      "epoch": 0.1527585452955324,
      "grad_norm": 19.504711044782532,
      "learning_rate": 9.612139775781766e-06,
      "loss": 1.1462,
      "step": 524
    },
    {
      "epoch": 0.1530500692369361,
      "grad_norm": 20.948578373015515,
      "learning_rate": 9.610315025185343e-06,
      "loss": 1.0547,
      "step": 525
    },
    {
      "epoch": 0.15334159317833976,
      "grad_norm": 39.123190749839004,
      "learning_rate": 9.608486166274444e-06,
      "loss": 1.1456,
      "step": 526
    },
    {
      "epoch": 0.15363311711974345,
      "grad_norm": 71.72105956473257,
      "learning_rate": 9.606653200678789e-06,
      "loss": 1.1524,
      "step": 527
    },
    {
      "epoch": 0.15392464106114714,
      "grad_norm": 46.8360644779971,
      "learning_rate": 9.60481613003176e-06,
      "loss": 0.9807,
      "step": 528
    },
    {
      "epoch": 0.15421616500255084,
      "grad_norm": 27.281875433773603,
      "learning_rate": 9.602974955970393e-06,
      "loss": 1.1492,
      "step": 529
    },
    {
      "epoch": 0.15450768894395453,
      "grad_norm": 52.183091589064304,
      "learning_rate": 9.601129680135386e-06,
      "loss": 1.1333,
      "step": 530
    },
    {
      "epoch": 0.15479921288535822,
      "grad_norm": 25.841569076692235,
      "learning_rate": 9.599280304171088e-06,
      "loss": 1.171,
      "step": 531
    },
    {
      "epoch": 0.1550907368267619,
      "grad_norm": 20.329215403543273,
      "learning_rate": 9.597426829725504e-06,
      "loss": 0.9597,
      "step": 532
    },
    {
      "epoch": 0.15538226076816558,
      "grad_norm": 18.400652977379753,
      "learning_rate": 9.59556925845029e-06,
      "loss": 1.1263,
      "step": 533
    },
    {
      "epoch": 0.15567378470956927,
      "grad_norm": 24.3690746454284,
      "learning_rate": 9.593707592000751e-06,
      "loss": 1.0623,
      "step": 534
    },
    {
      "epoch": 0.15596530865097297,
      "grad_norm": 21.872318818440508,
      "learning_rate": 9.591841832035847e-06,
      "loss": 1.191,
      "step": 535
    },
    {
      "epoch": 0.15625683259237666,
      "grad_norm": 301.1207317925737,
      "learning_rate": 9.58997198021818e-06,
      "loss": 1.1098,
      "step": 536
    },
    {
      "epoch": 0.15654835653378033,
      "grad_norm": 17.58425546683188,
      "learning_rate": 9.588098038214002e-06,
      "loss": 1.062,
      "step": 537
    },
    {
      "epoch": 0.15683988047518402,
      "grad_norm": 56.730357994492316,
      "learning_rate": 9.586220007693205e-06,
      "loss": 1.1917,
      "step": 538
    },
    {
      "epoch": 0.1571314044165877,
      "grad_norm": 23.163936187898777,
      "learning_rate": 9.58433789032933e-06,
      "loss": 0.9158,
      "step": 539
    },
    {
      "epoch": 0.1574229283579914,
      "grad_norm": 142.60266883232237,
      "learning_rate": 9.582451687799557e-06,
      "loss": 1.3131,
      "step": 540
    },
    {
      "epoch": 0.1577144522993951,
      "grad_norm": 54.47752185537803,
      "learning_rate": 9.580561401784706e-06,
      "loss": 1.1618,
      "step": 541
    },
    {
      "epoch": 0.15800597624079876,
      "grad_norm": 22.747223833917886,
      "learning_rate": 9.578667033969238e-06,
      "loss": 1.0117,
      "step": 542
    },
    {
      "epoch": 0.15829750018220246,
      "grad_norm": 31.30252510260865,
      "learning_rate": 9.576768586041248e-06,
      "loss": 1.0904,
      "step": 543
    },
    {
      "epoch": 0.15858902412360615,
      "grad_norm": 32.81424641815846,
      "learning_rate": 9.574866059692471e-06,
      "loss": 1.043,
      "step": 544
    },
    {
      "epoch": 0.15888054806500984,
      "grad_norm": 50.661787112191405,
      "learning_rate": 9.572959456618273e-06,
      "loss": 1.0762,
      "step": 545
    },
    {
      "epoch": 0.15917207200641353,
      "grad_norm": 25.58243738602507,
      "learning_rate": 9.571048778517655e-06,
      "loss": 1.1371,
      "step": 546
    },
    {
      "epoch": 0.15946359594781723,
      "grad_norm": 13.38726688426862,
      "learning_rate": 9.569134027093246e-06,
      "loss": 0.8974,
      "step": 547
    },
    {
      "epoch": 0.1597551198892209,
      "grad_norm": 34.79940047185494,
      "learning_rate": 9.567215204051307e-06,
      "loss": 1.202,
      "step": 548
    },
    {
      "epoch": 0.16004664383062459,
      "grad_norm": 73.9451325573122,
      "learning_rate": 9.565292311101731e-06,
      "loss": 1.2328,
      "step": 549
    },
    {
      "epoch": 0.16033816777202828,
      "grad_norm": 21.904871105014305,
      "learning_rate": 9.563365349958032e-06,
      "loss": 1.1157,
      "step": 550
    },
    {
      "epoch": 0.16033816777202828,
      "eval_loss": 1.1784170866012573,
      "eval_runtime": 1919.2021,
      "eval_samples_per_second": 8.999,
      "eval_steps_per_second": 0.5,
      "step": 550
    },
    {
      "epoch": 0.16062969171343197,
      "grad_norm": 15.740271006854272,
      "learning_rate": 9.561434322337349e-06,
      "loss": 1.1025,
      "step": 551
    },
    {
      "epoch": 0.16092121565483566,
      "grad_norm": 32.8412088022258,
      "learning_rate": 9.55949922996045e-06,
      "loss": 0.7926,
      "step": 552
    },
    {
      "epoch": 0.16121273959623933,
      "grad_norm": 57.18091961708963,
      "learning_rate": 9.557560074551722e-06,
      "loss": 1.368,
      "step": 553
    },
    {
      "epoch": 0.16150426353764302,
      "grad_norm": 28.8106532707237,
      "learning_rate": 9.555616857839171e-06,
      "loss": 1.0502,
      "step": 554
    },
    {
      "epoch": 0.16179578747904672,
      "grad_norm": 50.9925111885848,
      "learning_rate": 9.553669581554427e-06,
      "loss": 0.9284,
      "step": 555
    },
    {
      "epoch": 0.1620873114204504,
      "grad_norm": 19.751894118624527,
      "learning_rate": 9.551718247432732e-06,
      "loss": 1.0995,
      "step": 556
    },
    {
      "epoch": 0.1623788353618541,
      "grad_norm": 37.798999262157636,
      "learning_rate": 9.549762857212944e-06,
      "loss": 1.1231,
      "step": 557
    },
    {
      "epoch": 0.16267035930325777,
      "grad_norm": 34.08393238287175,
      "learning_rate": 9.547803412637542e-06,
      "loss": 1.1078,
      "step": 558
    },
    {
      "epoch": 0.16296188324466146,
      "grad_norm": 22.213200665136405,
      "learning_rate": 9.545839915452612e-06,
      "loss": 0.9837,
      "step": 559
    },
    {
      "epoch": 0.16325340718606515,
      "grad_norm": 47.79641112852825,
      "learning_rate": 9.543872367407854e-06,
      "loss": 0.8886,
      "step": 560
    },
    {
      "epoch": 0.16354493112746885,
      "grad_norm": 25.352128450707607,
      "learning_rate": 9.541900770256576e-06,
      "loss": 1.0266,
      "step": 561
    },
    {
      "epoch": 0.16383645506887254,
      "grad_norm": 19.008802432527464,
      "learning_rate": 9.539925125755695e-06,
      "loss": 1.0207,
      "step": 562
    },
    {
      "epoch": 0.16412797901027623,
      "grad_norm": 24.792734519167144,
      "learning_rate": 9.537945435665735e-06,
      "loss": 0.9742,
      "step": 563
    },
    {
      "epoch": 0.1644195029516799,
      "grad_norm": 21.079753317277,
      "learning_rate": 9.535961701750825e-06,
      "loss": 0.974,
      "step": 564
    },
    {
      "epoch": 0.1647110268930836,
      "grad_norm": 37.93741399058735,
      "learning_rate": 9.533973925778697e-06,
      "loss": 1.237,
      "step": 565
    },
    {
      "epoch": 0.16500255083448728,
      "grad_norm": 92.55181064238714,
      "learning_rate": 9.531982109520686e-06,
      "loss": 1.142,
      "step": 566
    },
    {
      "epoch": 0.16529407477589098,
      "grad_norm": 15.21604553312173,
      "learning_rate": 9.529986254751724e-06,
      "loss": 0.9506,
      "step": 567
    },
    {
      "epoch": 0.16558559871729467,
      "grad_norm": 69.28798170433903,
      "learning_rate": 9.527986363250348e-06,
      "loss": 1.2251,
      "step": 568
    },
    {
      "epoch": 0.16587712265869833,
      "grad_norm": 44.306836913241746,
      "learning_rate": 9.525982436798684e-06,
      "loss": 1.0757,
      "step": 569
    },
    {
      "epoch": 0.16616864660010203,
      "grad_norm": 252.47296630877312,
      "learning_rate": 9.523974477182465e-06,
      "loss": 1.0144,
      "step": 570
    },
    {
      "epoch": 0.16646017054150572,
      "grad_norm": 71.45891106137577,
      "learning_rate": 9.521962486191004e-06,
      "loss": 1.1692,
      "step": 571
    },
    {
      "epoch": 0.1667516944829094,
      "grad_norm": 27.05347339065216,
      "learning_rate": 9.519946465617217e-06,
      "loss": 1.0174,
      "step": 572
    },
    {
      "epoch": 0.1670432184243131,
      "grad_norm": 26.195299846466522,
      "learning_rate": 9.517926417257609e-06,
      "loss": 1.1463,
      "step": 573
    },
    {
      "epoch": 0.16733474236571677,
      "grad_norm": 23.364633858926016,
      "learning_rate": 9.515902342912268e-06,
      "loss": 1.0702,
      "step": 574
    },
    {
      "epoch": 0.16762626630712046,
      "grad_norm": 22.43844088163322,
      "learning_rate": 9.513874244384878e-06,
      "loss": 1.0297,
      "step": 575
    },
    {
      "epoch": 0.16791779024852416,
      "grad_norm": 38.300010519903225,
      "learning_rate": 9.511842123482703e-06,
      "loss": 1.1462,
      "step": 576
    },
    {
      "epoch": 0.16820931418992785,
      "grad_norm": 25.13582386859589,
      "learning_rate": 9.509805982016596e-06,
      "loss": 1.087,
      "step": 577
    },
    {
      "epoch": 0.16850083813133154,
      "grad_norm": 18.760482146556363,
      "learning_rate": 9.507765821800988e-06,
      "loss": 1.1587,
      "step": 578
    },
    {
      "epoch": 0.16879236207273524,
      "grad_norm": 43.8136690472008,
      "learning_rate": 9.505721644653893e-06,
      "loss": 1.0806,
      "step": 579
    },
    {
      "epoch": 0.1690838860141389,
      "grad_norm": 31.11255809975462,
      "learning_rate": 9.503673452396909e-06,
      "loss": 1.0508,
      "step": 580
    },
    {
      "epoch": 0.1693754099555426,
      "grad_norm": 39.298275570291246,
      "learning_rate": 9.501621246855206e-06,
      "loss": 1.1278,
      "step": 581
    },
    {
      "epoch": 0.1696669338969463,
      "grad_norm": 19.815229782817408,
      "learning_rate": 9.499565029857529e-06,
      "loss": 1.2254,
      "step": 582
    },
    {
      "epoch": 0.16995845783834998,
      "grad_norm": 23.72314721698866,
      "learning_rate": 9.497504803236207e-06,
      "loss": 1.0306,
      "step": 583
    },
    {
      "epoch": 0.17024998177975367,
      "grad_norm": 24.178436919905877,
      "learning_rate": 9.49544056882713e-06,
      "loss": 0.9639,
      "step": 584
    },
    {
      "epoch": 0.17054150572115734,
      "grad_norm": 37.629643414150486,
      "learning_rate": 9.49337232846977e-06,
      "loss": 0.9653,
      "step": 585
    },
    {
      "epoch": 0.17083302966256103,
      "grad_norm": 34.10620383303592,
      "learning_rate": 9.491300084007162e-06,
      "loss": 0.8491,
      "step": 586
    },
    {
      "epoch": 0.17112455360396472,
      "grad_norm": 17.93355691157691,
      "learning_rate": 9.48922383728591e-06,
      "loss": 1.1788,
      "step": 587
    },
    {
      "epoch": 0.17141607754536842,
      "grad_norm": 69.80639237807418,
      "learning_rate": 9.48714359015619e-06,
      "loss": 1.0183,
      "step": 588
    },
    {
      "epoch": 0.1717076014867721,
      "grad_norm": 15.741317614381765,
      "learning_rate": 9.485059344471735e-06,
      "loss": 1.2029,
      "step": 589
    },
    {
      "epoch": 0.17199912542817578,
      "grad_norm": 17.93447614442211,
      "learning_rate": 9.482971102089845e-06,
      "loss": 0.963,
      "step": 590
    },
    {
      "epoch": 0.17229064936957947,
      "grad_norm": 22.333276646318083,
      "learning_rate": 9.48087886487138e-06,
      "loss": 1.0434,
      "step": 591
    },
    {
      "epoch": 0.17258217331098316,
      "grad_norm": 34.433970423272854,
      "learning_rate": 9.478782634680765e-06,
      "loss": 1.0838,
      "step": 592
    },
    {
      "epoch": 0.17287369725238685,
      "grad_norm": 23.784700171240168,
      "learning_rate": 9.476682413385978e-06,
      "loss": 1.1435,
      "step": 593
    },
    {
      "epoch": 0.17316522119379055,
      "grad_norm": 32.35024221732807,
      "learning_rate": 9.47457820285855e-06,
      "loss": 1.0512,
      "step": 594
    },
    {
      "epoch": 0.17345674513519424,
      "grad_norm": 28.002761422381564,
      "learning_rate": 9.47247000497358e-06,
      "loss": 0.9677,
      "step": 595
    },
    {
      "epoch": 0.1737482690765979,
      "grad_norm": 31.21066711152469,
      "learning_rate": 9.470357821609703e-06,
      "loss": 1.1965,
      "step": 596
    },
    {
      "epoch": 0.1740397930180016,
      "grad_norm": 27.527082875255136,
      "learning_rate": 9.468241654649121e-06,
      "loss": 1.0581,
      "step": 597
    },
    {
      "epoch": 0.1743313169594053,
      "grad_norm": 42.87429662890905,
      "learning_rate": 9.466121505977577e-06,
      "loss": 1.1113,
      "step": 598
    },
    {
      "epoch": 0.17462284090080898,
      "grad_norm": 27.35635781843465,
      "learning_rate": 9.463997377484364e-06,
      "loss": 1.0624,
      "step": 599
    },
    {
      "epoch": 0.17491436484221268,
      "grad_norm": 49.78283614895045,
      "learning_rate": 9.461869271062322e-06,
      "loss": 0.9789,
      "step": 600
    },
    {
      "epoch": 0.17491436484221268,
      "eval_loss": 1.1793168783187866,
      "eval_runtime": 1914.2447,
      "eval_samples_per_second": 9.022,
      "eval_steps_per_second": 0.502,
      "step": 600
    },
    {
      "epoch": 0.17520588878361634,
      "grad_norm": 26.39403713635298,
      "learning_rate": 9.459737188607835e-06,
      "loss": 1.0583,
      "step": 601
    },
    {
      "epoch": 0.17549741272502004,
      "grad_norm": 30.58339912502805,
      "learning_rate": 9.457601132020832e-06,
      "loss": 1.2556,
      "step": 602
    },
    {
      "epoch": 0.17578893666642373,
      "grad_norm": 38.00542075823468,
      "learning_rate": 9.45546110320478e-06,
      "loss": 1.0166,
      "step": 603
    },
    {
      "epoch": 0.17608046060782742,
      "grad_norm": 26.185389492624843,
      "learning_rate": 9.453317104066693e-06,
      "loss": 1.1155,
      "step": 604
    },
    {
      "epoch": 0.17637198454923111,
      "grad_norm": 37.91277796870061,
      "learning_rate": 9.45116913651711e-06,
      "loss": 1.0091,
      "step": 605
    },
    {
      "epoch": 0.17666350849063478,
      "grad_norm": 21.289747832288295,
      "learning_rate": 9.44901720247012e-06,
      "loss": 1.1683,
      "step": 606
    },
    {
      "epoch": 0.17695503243203847,
      "grad_norm": 26.759523563190623,
      "learning_rate": 9.446861303843338e-06,
      "loss": 1.098,
      "step": 607
    },
    {
      "epoch": 0.17724655637344217,
      "grad_norm": 25.568832530645278,
      "learning_rate": 9.444701442557917e-06,
      "loss": 1.0745,
      "step": 608
    },
    {
      "epoch": 0.17753808031484586,
      "grad_norm": 49.54112096647874,
      "learning_rate": 9.442537620538535e-06,
      "loss": 1.159,
      "step": 609
    },
    {
      "epoch": 0.17782960425624955,
      "grad_norm": 38.04817585835372,
      "learning_rate": 9.440369839713407e-06,
      "loss": 1.0687,
      "step": 610
    },
    {
      "epoch": 0.17812112819765324,
      "grad_norm": 27.119928252721664,
      "learning_rate": 9.438198102014271e-06,
      "loss": 1.1631,
      "step": 611
    },
    {
      "epoch": 0.1784126521390569,
      "grad_norm": 90.92045865251326,
      "learning_rate": 9.436022409376391e-06,
      "loss": 1.2789,
      "step": 612
    },
    {
      "epoch": 0.1787041760804606,
      "grad_norm": 23.610681400372897,
      "learning_rate": 9.433842763738557e-06,
      "loss": 0.9668,
      "step": 613
    },
    {
      "epoch": 0.1789957000218643,
      "grad_norm": 36.50139474722115,
      "learning_rate": 9.431659167043079e-06,
      "loss": 1.0368,
      "step": 614
    },
    {
      "epoch": 0.179287223963268,
      "grad_norm": 19.714575812423853,
      "learning_rate": 9.429471621235794e-06,
      "loss": 1.0237,
      "step": 615
    },
    {
      "epoch": 0.17957874790467168,
      "grad_norm": 18.717405904998667,
      "learning_rate": 9.427280128266049e-06,
      "loss": 1.0056,
      "step": 616
    },
    {
      "epoch": 0.17987027184607535,
      "grad_norm": 29.121945937483236,
      "learning_rate": 9.42508469008672e-06,
      "loss": 1.1215,
      "step": 617
    },
    {
      "epoch": 0.18016179578747904,
      "grad_norm": 45.69791390442322,
      "learning_rate": 9.422885308654183e-06,
      "loss": 0.9792,
      "step": 618
    },
    {
      "epoch": 0.18045331972888273,
      "grad_norm": 31.014042363258422,
      "learning_rate": 9.420681985928344e-06,
      "loss": 1.0077,
      "step": 619
    },
    {
      "epoch": 0.18074484367028643,
      "grad_norm": 18.4133621964713,
      "learning_rate": 9.418474723872609e-06,
      "loss": 1.1058,
      "step": 620
    },
    {
      "epoch": 0.18103636761169012,
      "grad_norm": 28.760120135192306,
      "learning_rate": 9.4162635244539e-06,
      "loss": 1.0313,
      "step": 621
    },
    {
      "epoch": 0.18132789155309378,
      "grad_norm": 32.59524334483209,
      "learning_rate": 9.414048389642652e-06,
      "loss": 1.1847,
      "step": 622
    },
    {
      "epoch": 0.18161941549449748,
      "grad_norm": 18.30820327545191,
      "learning_rate": 9.411829321412796e-06,
      "loss": 0.9021,
      "step": 623
    },
    {
      "epoch": 0.18191093943590117,
      "grad_norm": 29.494376346802227,
      "learning_rate": 9.409606321741776e-06,
      "loss": 1.1383,
      "step": 624
    },
    {
      "epoch": 0.18220246337730486,
      "grad_norm": 16.365295943794735,
      "learning_rate": 9.407379392610537e-06,
      "loss": 1.0424,
      "step": 625
    },
    {
      "epoch": 0.18249398731870856,
      "grad_norm": 28.914036222871292,
      "learning_rate": 9.405148536003527e-06,
      "loss": 1.2092,
      "step": 626
    },
    {
      "epoch": 0.18278551126011225,
      "grad_norm": 20.828835067450257,
      "learning_rate": 9.402913753908691e-06,
      "loss": 1.0174,
      "step": 627
    },
    {
      "epoch": 0.18307703520151591,
      "grad_norm": 35.5065162068191,
      "learning_rate": 9.400675048317473e-06,
      "loss": 0.9358,
      "step": 628
    },
    {
      "epoch": 0.1833685591429196,
      "grad_norm": 28.253480393711047,
      "learning_rate": 9.39843242122482e-06,
      "loss": 1.0961,
      "step": 629
    },
    {
      "epoch": 0.1836600830843233,
      "grad_norm": 38.70546437090655,
      "learning_rate": 9.396185874629158e-06,
      "loss": 1.0286,
      "step": 630
    },
    {
      "epoch": 0.183951607025727,
      "grad_norm": 41.519956041478565,
      "learning_rate": 9.393935410532421e-06,
      "loss": 1.0234,
      "step": 631
    },
    {
      "epoch": 0.18424313096713069,
      "grad_norm": 17.899722378946322,
      "learning_rate": 9.391681030940031e-06,
      "loss": 0.8194,
      "step": 632
    },
    {
      "epoch": 0.18453465490853435,
      "grad_norm": 19.915303749643616,
      "learning_rate": 9.38942273786089e-06,
      "loss": 0.9846,
      "step": 633
    },
    {
      "epoch": 0.18482617884993804,
      "grad_norm": 47.34944230712801,
      "learning_rate": 9.387160533307398e-06,
      "loss": 1.0766,
      "step": 634
    },
    {
      "epoch": 0.18511770279134174,
      "grad_norm": 28.93390687195922,
      "learning_rate": 9.384894419295433e-06,
      "loss": 0.937,
      "step": 635
    },
    {
      "epoch": 0.18540922673274543,
      "grad_norm": 53.106740235583466,
      "learning_rate": 9.382624397844363e-06,
      "loss": 1.0532,
      "step": 636
    },
    {
      "epoch": 0.18570075067414912,
      "grad_norm": 16.145931617205,
      "learning_rate": 9.380350470977033e-06,
      "loss": 1.0353,
      "step": 637
    },
    {
      "epoch": 0.1859922746155528,
      "grad_norm": 16.101508127383262,
      "learning_rate": 9.378072640719773e-06,
      "loss": 0.9897,
      "step": 638
    },
    {
      "epoch": 0.18628379855695648,
      "grad_norm": 52.1441760376116,
      "learning_rate": 9.375790909102385e-06,
      "loss": 1.1319,
      "step": 639
    },
    {
      "epoch": 0.18657532249836017,
      "grad_norm": 19.62124030266703,
      "learning_rate": 9.373505278158152e-06,
      "loss": 0.9398,
      "step": 640
    },
    {
      "epoch": 0.18686684643976387,
      "grad_norm": 15.673782364293727,
      "learning_rate": 9.371215749923834e-06,
      "loss": 1.153,
      "step": 641
    },
    {
      "epoch": 0.18715837038116756,
      "grad_norm": 25.05714124734551,
      "learning_rate": 9.368922326439655e-06,
      "loss": 1.0716,
      "step": 642
    },
    {
      "epoch": 0.18744989432257125,
      "grad_norm": 23.796729901734615,
      "learning_rate": 9.366625009749322e-06,
      "loss": 1.0053,
      "step": 643
    },
    {
      "epoch": 0.18774141826397492,
      "grad_norm": 41.42327061532633,
      "learning_rate": 9.364323801900002e-06,
      "loss": 1.0044,
      "step": 644
    },
    {
      "epoch": 0.1880329422053786,
      "grad_norm": 45.82211950938244,
      "learning_rate": 9.362018704942332e-06,
      "loss": 1.099,
      "step": 645
    },
    {
      "epoch": 0.1883244661467823,
      "grad_norm": 19.689786959258647,
      "learning_rate": 9.359709720930417e-06,
      "loss": 0.8684,
      "step": 646
    },
    {
      "epoch": 0.188615990088186,
      "grad_norm": 24.02489815652354,
      "learning_rate": 9.357396851921824e-06,
      "loss": 1.184,
      "step": 647
    },
    {
      "epoch": 0.1889075140295897,
      "grad_norm": 31.80554992431227,
      "learning_rate": 9.355080099977579e-06,
      "loss": 1.076,
      "step": 648
    },
    {
      "epoch": 0.18919903797099336,
      "grad_norm": 17.816876855271694,
      "learning_rate": 9.352759467162175e-06,
      "loss": 0.9055,
      "step": 649
    },
    {
      "epoch": 0.18949056191239705,
      "grad_norm": 37.35393186842933,
      "learning_rate": 9.350434955543557e-06,
      "loss": 0.9827,
      "step": 650
    },
    {
      "epoch": 0.18949056191239705,
      "eval_loss": 1.1714248657226562,
      "eval_runtime": 2107.0282,
      "eval_samples_per_second": 8.196,
      "eval_steps_per_second": 0.456,
      "step": 650
    },
    {
      "epoch": 0.18978208585380074,
      "grad_norm": 24.617581056743028,
      "learning_rate": 9.348106567193132e-06,
      "loss": 1.1021,
      "step": 651
    },
    {
      "epoch": 0.19007360979520443,
      "grad_norm": 13.182678578619896,
      "learning_rate": 9.345774304185756e-06,
      "loss": 0.8638,
      "step": 652
    },
    {
      "epoch": 0.19036513373660813,
      "grad_norm": 44.0695288338105,
      "learning_rate": 9.343438168599741e-06,
      "loss": 1.1541,
      "step": 653
    },
    {
      "epoch": 0.19065665767801182,
      "grad_norm": 36.17570388542415,
      "learning_rate": 9.341098162516848e-06,
      "loss": 1.0185,
      "step": 654
    },
    {
      "epoch": 0.19094818161941549,
      "grad_norm": 34.33408201967094,
      "learning_rate": 9.338754288022292e-06,
      "loss": 1.068,
      "step": 655
    },
    {
      "epoch": 0.19123970556081918,
      "grad_norm": 18.5634752033769,
      "learning_rate": 9.336406547204726e-06,
      "loss": 0.9927,
      "step": 656
    },
    {
      "epoch": 0.19153122950222287,
      "grad_norm": 94.17111585704093,
      "learning_rate": 9.334054942156258e-06,
      "loss": 1.0311,
      "step": 657
    },
    {
      "epoch": 0.19182275344362656,
      "grad_norm": 28.777978722231435,
      "learning_rate": 9.331699474972434e-06,
      "loss": 1.0497,
      "step": 658
    },
    {
      "epoch": 0.19211427738503026,
      "grad_norm": 43.71962326043472,
      "learning_rate": 9.329340147752243e-06,
      "loss": 1.0253,
      "step": 659
    },
    {
      "epoch": 0.19240580132643392,
      "grad_norm": 40.99477417523327,
      "learning_rate": 9.326976962598113e-06,
      "loss": 1.0046,
      "step": 660
    },
    {
      "epoch": 0.19269732526783762,
      "grad_norm": 69.02591502386703,
      "learning_rate": 9.324609921615911e-06,
      "loss": 1.0972,
      "step": 661
    },
    {
      "epoch": 0.1929888492092413,
      "grad_norm": 88.22462394389738,
      "learning_rate": 9.322239026914938e-06,
      "loss": 0.9619,
      "step": 662
    },
    {
      "epoch": 0.193280373150645,
      "grad_norm": 45.02927005643994,
      "learning_rate": 9.319864280607935e-06,
      "loss": 0.8915,
      "step": 663
    },
    {
      "epoch": 0.1935718970920487,
      "grad_norm": 28.65231623272117,
      "learning_rate": 9.317485684811065e-06,
      "loss": 1.0149,
      "step": 664
    },
    {
      "epoch": 0.19386342103345236,
      "grad_norm": 100.62996800234934,
      "learning_rate": 9.31510324164393e-06,
      "loss": 1.0873,
      "step": 665
    },
    {
      "epoch": 0.19415494497485605,
      "grad_norm": 29.15015470494994,
      "learning_rate": 9.31271695322956e-06,
      "loss": 1.0546,
      "step": 666
    },
    {
      "epoch": 0.19444646891625975,
      "grad_norm": 25.027977556277648,
      "learning_rate": 9.310326821694403e-06,
      "loss": 0.9183,
      "step": 667
    },
    {
      "epoch": 0.19473799285766344,
      "grad_norm": 137.8621158084867,
      "learning_rate": 9.307932849168341e-06,
      "loss": 1.1869,
      "step": 668
    },
    {
      "epoch": 0.19502951679906713,
      "grad_norm": 22.237869825399244,
      "learning_rate": 9.30553503778468e-06,
      "loss": 1.0832,
      "step": 669
    },
    {
      "epoch": 0.19532104074047082,
      "grad_norm": 30.177970837587146,
      "learning_rate": 9.303133389680134e-06,
      "loss": 1.0194,
      "step": 670
    },
    {
      "epoch": 0.1956125646818745,
      "grad_norm": 31.52787037288531,
      "learning_rate": 9.300727906994848e-06,
      "loss": 1.0218,
      "step": 671
    },
    {
      "epoch": 0.19590408862327818,
      "grad_norm": 35.37618244492424,
      "learning_rate": 9.298318591872381e-06,
      "loss": 0.9351,
      "step": 672
    },
    {
      "epoch": 0.19619561256468188,
      "grad_norm": 23.226425591636367,
      "learning_rate": 9.295905446459708e-06,
      "loss": 0.9966,
      "step": 673
    },
    {
      "epoch": 0.19648713650608557,
      "grad_norm": 34.194846270543415,
      "learning_rate": 9.293488472907213e-06,
      "loss": 0.9224,
      "step": 674
    },
    {
      "epoch": 0.19677866044748926,
      "grad_norm": 28.766615627327905,
      "learning_rate": 9.291067673368694e-06,
      "loss": 0.8947,
      "step": 675
    },
    {
      "epoch": 0.19707018438889293,
      "grad_norm": 28.880352503220106,
      "learning_rate": 9.288643050001362e-06,
      "loss": 1.0109,
      "step": 676
    },
    {
      "epoch": 0.19736170833029662,
      "grad_norm": 23.503129577397488,
      "learning_rate": 9.286214604965826e-06,
      "loss": 1.0397,
      "step": 677
    },
    {
      "epoch": 0.1976532322717003,
      "grad_norm": 66.14556042623057,
      "learning_rate": 9.283782340426112e-06,
      "loss": 0.9904,
      "step": 678
    },
    {
      "epoch": 0.197944756213104,
      "grad_norm": 48.37657703698637,
      "learning_rate": 9.281346258549639e-06,
      "loss": 1.0844,
      "step": 679
    },
    {
      "epoch": 0.1982362801545077,
      "grad_norm": 17.47190048122972,
      "learning_rate": 9.278906361507238e-06,
      "loss": 1.0918,
      "step": 680
    },
    {
      "epoch": 0.19852780409591136,
      "grad_norm": 41.7355470347729,
      "learning_rate": 9.27646265147313e-06,
      "loss": 1.0511,
      "step": 681
    },
    {
      "epoch": 0.19881932803731506,
      "grad_norm": 29.855879764241404,
      "learning_rate": 9.274015130624943e-06,
      "loss": 1.1927,
      "step": 682
    },
    {
      "epoch": 0.19911085197871875,
      "grad_norm": 35.17521135664104,
      "learning_rate": 9.27156380114369e-06,
      "loss": 0.9878,
      "step": 683
    },
    {
      "epoch": 0.19940237592012244,
      "grad_norm": 72.11600222379535,
      "learning_rate": 9.26910866521379e-06,
      "loss": 0.8952,
      "step": 684
    },
    {
      "epoch": 0.19969389986152614,
      "grad_norm": 22.079062307543666,
      "learning_rate": 9.266649725023045e-06,
      "loss": 1.0654,
      "step": 685
    },
    {
      "epoch": 0.19998542380292983,
      "grad_norm": 47.28424885704714,
      "learning_rate": 9.264186982762649e-06,
      "loss": 1.0945,
      "step": 686
    },
    {
      "epoch": 0.2002769477443335,
      "grad_norm": 25.010565324747855,
      "learning_rate": 9.261720440627187e-06,
      "loss": 0.7919,
      "step": 687
    },
    {
      "epoch": 0.2005684716857372,
      "grad_norm": 49.29630020038456,
      "learning_rate": 9.25925010081463e-06,
      "loss": 1.1045,
      "step": 688
    },
    {
      "epoch": 0.20085999562714088,
      "grad_norm": 25.42367013276608,
      "learning_rate": 9.256775965526327e-06,
      "loss": 0.9626,
      "step": 689
    },
    {
      "epoch": 0.20115151956854457,
      "grad_norm": 24.28309370919054,
      "learning_rate": 9.254298036967015e-06,
      "loss": 1.0762,
      "step": 690
    },
    {
      "epoch": 0.20144304350994827,
      "grad_norm": 40.95459765726716,
      "learning_rate": 9.251816317344814e-06,
      "loss": 1.1253,
      "step": 691
    },
    {
      "epoch": 0.20173456745135193,
      "grad_norm": 21.306242276852284,
      "learning_rate": 9.249330808871213e-06,
      "loss": 0.9974,
      "step": 692
    },
    {
      "epoch": 0.20202609139275562,
      "grad_norm": 26.48900190100961,
      "learning_rate": 9.246841513761086e-06,
      "loss": 0.9611,
      "step": 693
    },
    {
      "epoch": 0.20231761533415932,
      "grad_norm": 25.122711854814167,
      "learning_rate": 9.244348434232676e-06,
      "loss": 1.2076,
      "step": 694
    },
    {
      "epoch": 0.202609139275563,
      "grad_norm": 45.86188649736369,
      "learning_rate": 9.241851572507599e-06,
      "loss": 0.9364,
      "step": 695
    },
    {
      "epoch": 0.2029006632169667,
      "grad_norm": 27.25538432250588,
      "learning_rate": 9.239350930810843e-06,
      "loss": 1.0282,
      "step": 696
    },
    {
      "epoch": 0.20319218715837037,
      "grad_norm": 51.75267420282361,
      "learning_rate": 9.236846511370767e-06,
      "loss": 0.9312,
      "step": 697
    },
    {
      "epoch": 0.20348371109977406,
      "grad_norm": 55.66032045744652,
      "learning_rate": 9.23433831641909e-06,
      "loss": 1.1065,
      "step": 698
    },
    {
      "epoch": 0.20377523504117775,
      "grad_norm": 42.136563111984366,
      "learning_rate": 9.231826348190897e-06,
      "loss": 0.8592,
      "step": 699
    },
    {
      "epoch": 0.20406675898258145,
      "grad_norm": 22.618797130036818,
      "learning_rate": 9.229310608924643e-06,
      "loss": 0.9536,
      "step": 700
    },
    {
      "epoch": 0.20406675898258145,
      "eval_loss": 1.1717129945755005,
      "eval_runtime": 2087.4592,
      "eval_samples_per_second": 8.273,
      "eval_steps_per_second": 0.46,
      "step": 700
    },
    {
      "epoch": 0.20435828292398514,
      "grad_norm": 42.103619784063056,
      "learning_rate": 9.226791100862134e-06,
      "loss": 0.8207,
      "step": 701
    },
    {
      "epoch": 0.20464980686538883,
      "grad_norm": 33.55334426749062,
      "learning_rate": 9.224267826248536e-06,
      "loss": 1.1624,
      "step": 702
    },
    {
      "epoch": 0.2049413308067925,
      "grad_norm": 30.885141851545974,
      "learning_rate": 9.221740787332376e-06,
      "loss": 1.3188,
      "step": 703
    },
    {
      "epoch": 0.2052328547481962,
      "grad_norm": 33.33432015072711,
      "learning_rate": 9.219209986365533e-06,
      "loss": 0.9763,
      "step": 704
    },
    {
      "epoch": 0.20552437868959988,
      "grad_norm": 25.452839618678045,
      "learning_rate": 9.216675425603237e-06,
      "loss": 1.0142,
      "step": 705
    },
    {
      "epoch": 0.20581590263100358,
      "grad_norm": 20.410817750179092,
      "learning_rate": 9.21413710730407e-06,
      "loss": 0.8973,
      "step": 706
    },
    {
      "epoch": 0.20610742657240727,
      "grad_norm": 31.158719298527977,
      "learning_rate": 9.211595033729963e-06,
      "loss": 0.8673,
      "step": 707
    },
    {
      "epoch": 0.20639895051381094,
      "grad_norm": 29.408161443627478,
      "learning_rate": 9.20904920714619e-06,
      "loss": 0.985,
      "step": 708
    },
    {
      "epoch": 0.20669047445521463,
      "grad_norm": 39.604644161683076,
      "learning_rate": 9.206499629821373e-06,
      "loss": 1.0051,
      "step": 709
    },
    {
      "epoch": 0.20698199839661832,
      "grad_norm": 22.737565140962346,
      "learning_rate": 9.203946304027476e-06,
      "loss": 1.0659,
      "step": 710
    },
    {
      "epoch": 0.20727352233802201,
      "grad_norm": 51.7999137670232,
      "learning_rate": 9.201389232039802e-06,
      "loss": 0.9802,
      "step": 711
    },
    {
      "epoch": 0.2075650462794257,
      "grad_norm": 36.12511964751788,
      "learning_rate": 9.198828416136991e-06,
      "loss": 0.9892,
      "step": 712
    },
    {
      "epoch": 0.20785657022082937,
      "grad_norm": 53.98959588773419,
      "learning_rate": 9.196263858601023e-06,
      "loss": 0.9463,
      "step": 713
    },
    {
      "epoch": 0.20814809416223307,
      "grad_norm": 27.32423671911781,
      "learning_rate": 9.193695561717207e-06,
      "loss": 0.9238,
      "step": 714
    },
    {
      "epoch": 0.20843961810363676,
      "grad_norm": 33.93350471169292,
      "learning_rate": 9.19112352777419e-06,
      "loss": 1.0648,
      "step": 715
    },
    {
      "epoch": 0.20873114204504045,
      "grad_norm": 136.6665309029085,
      "learning_rate": 9.188547759063948e-06,
      "loss": 0.9883,
      "step": 716
    },
    {
      "epoch": 0.20902266598644414,
      "grad_norm": 19.274888281635622,
      "learning_rate": 9.185968257881782e-06,
      "loss": 1.0809,
      "step": 717
    },
    {
      "epoch": 0.20931418992784784,
      "grad_norm": 21.561511130204543,
      "learning_rate": 9.183385026526317e-06,
      "loss": 1.0823,
      "step": 718
    },
    {
      "epoch": 0.2096057138692515,
      "grad_norm": 32.60770035958628,
      "learning_rate": 9.18079806729951e-06,
      "loss": 0.9531,
      "step": 719
    },
    {
      "epoch": 0.2098972378106552,
      "grad_norm": 33.02611774543217,
      "learning_rate": 9.178207382506634e-06,
      "loss": 0.9575,
      "step": 720
    },
    {
      "epoch": 0.2101887617520589,
      "grad_norm": 120.54986998511629,
      "learning_rate": 9.175612974456284e-06,
      "loss": 1.044,
      "step": 721
    },
    {
      "epoch": 0.21048028569346258,
      "grad_norm": 34.03876669049837,
      "learning_rate": 9.173014845460375e-06,
      "loss": 1.2201,
      "step": 722
    },
    {
      "epoch": 0.21077180963486628,
      "grad_norm": 22.934926963591945,
      "learning_rate": 9.17041299783413e-06,
      "loss": 1.0313,
      "step": 723
    },
    {
      "epoch": 0.21106333357626994,
      "grad_norm": 29.647744833395002,
      "learning_rate": 9.167807433896091e-06,
      "loss": 0.9185,
      "step": 724
    },
    {
      "epoch": 0.21135485751767363,
      "grad_norm": 17.65070089465046,
      "learning_rate": 9.165198155968118e-06,
      "loss": 1.1067,
      "step": 725
    },
    {
      "epoch": 0.21164638145907733,
      "grad_norm": 33.31110045596932,
      "learning_rate": 9.162585166375367e-06,
      "loss": 0.9491,
      "step": 726
    },
    {
      "epoch": 0.21193790540048102,
      "grad_norm": 64.34186725649703,
      "learning_rate": 9.159968467446312e-06,
      "loss": 0.9083,
      "step": 727
    },
    {
      "epoch": 0.2122294293418847,
      "grad_norm": 37.71885867417863,
      "learning_rate": 9.157348061512728e-06,
      "loss": 1.1248,
      "step": 728
    },
    {
      "epoch": 0.21252095328328838,
      "grad_norm": 26.659676447779773,
      "learning_rate": 9.154723950909694e-06,
      "loss": 1.1766,
      "step": 729
    },
    {
      "epoch": 0.21281247722469207,
      "grad_norm": 45.03160880485418,
      "learning_rate": 9.152096137975593e-06,
      "loss": 0.9912,
      "step": 730
    },
    {
      "epoch": 0.21310400116609576,
      "grad_norm": 24.87789147538287,
      "learning_rate": 9.1494646250521e-06,
      "loss": 1.134,
      "step": 731
    },
    {
      "epoch": 0.21339552510749946,
      "grad_norm": 12.744436624903726,
      "learning_rate": 9.146829414484198e-06,
      "loss": 1.0148,
      "step": 732
    },
    {
      "epoch": 0.21368704904890315,
      "grad_norm": 46.454415391959024,
      "learning_rate": 9.144190508620156e-06,
      "loss": 1.0195,
      "step": 733
    },
    {
      "epoch": 0.21397857299030684,
      "grad_norm": 16.163636429580876,
      "learning_rate": 9.14154790981154e-06,
      "loss": 0.8824,
      "step": 734
    },
    {
      "epoch": 0.2142700969317105,
      "grad_norm": 24.478710604605254,
      "learning_rate": 9.138901620413203e-06,
      "loss": 1.1361,
      "step": 735
    },
    {
      "epoch": 0.2145616208731142,
      "grad_norm": 27.988925691343713,
      "learning_rate": 9.136251642783294e-06,
      "loss": 0.9424,
      "step": 736
    },
    {
      "epoch": 0.2148531448145179,
      "grad_norm": 21.4342437350518,
      "learning_rate": 9.133597979283239e-06,
      "loss": 0.9894,
      "step": 737
    },
    {
      "epoch": 0.2151446687559216,
      "grad_norm": 24.389285105957036,
      "learning_rate": 9.130940632277757e-06,
      "loss": 0.8393,
      "step": 738
    },
    {
      "epoch": 0.21543619269732528,
      "grad_norm": 29.676878570668645,
      "learning_rate": 9.128279604134846e-06,
      "loss": 1.072,
      "step": 739
    },
    {
      "epoch": 0.21572771663872894,
      "grad_norm": 19.12064983050799,
      "learning_rate": 9.125614897225785e-06,
      "loss": 1.0456,
      "step": 740
    },
    {
      "epoch": 0.21601924058013264,
      "grad_norm": 17.20860780933045,
      "learning_rate": 9.122946513925128e-06,
      "loss": 1.0569,
      "step": 741
    },
    {
      "epoch": 0.21631076452153633,
      "grad_norm": 59.75790113375134,
      "learning_rate": 9.120274456610708e-06,
      "loss": 1.1948,
      "step": 742
    },
    {
      "epoch": 0.21660228846294002,
      "grad_norm": 29.64277737477631,
      "learning_rate": 9.117598727663635e-06,
      "loss": 1.0228,
      "step": 743
    },
    {
      "epoch": 0.21689381240434372,
      "grad_norm": 19.600080111779647,
      "learning_rate": 9.114919329468283e-06,
      "loss": 0.9225,
      "step": 744
    },
    {
      "epoch": 0.21718533634574738,
      "grad_norm": 19.994112779804404,
      "learning_rate": 9.112236264412303e-06,
      "loss": 1.0517,
      "step": 745
    },
    {
      "epoch": 0.21747686028715107,
      "grad_norm": 35.66546817548262,
      "learning_rate": 9.10954953488661e-06,
      "loss": 1.1846,
      "step": 746
    },
    {
      "epoch": 0.21776838422855477,
      "grad_norm": 24.167538973622054,
      "learning_rate": 9.106859143285386e-06,
      "loss": 0.9922,
      "step": 747
    },
    {
      "epoch": 0.21805990816995846,
      "grad_norm": 19.21150319267492,
      "learning_rate": 9.104165092006075e-06,
      "loss": 1.0392,
      "step": 748
    },
    {
      "epoch": 0.21835143211136215,
      "grad_norm": 35.928070810819385,
      "learning_rate": 9.101467383449383e-06,
      "loss": 1.1369,
      "step": 749
    },
    {
      "epoch": 0.21864295605276585,
      "grad_norm": 23.123207968401413,
      "learning_rate": 9.098766020019273e-06,
      "loss": 1.081,
      "step": 750
    },
    {
      "epoch": 0.21864295605276585,
      "eval_loss": 1.1692010164260864,
      "eval_runtime": 2063.7637,
      "eval_samples_per_second": 8.368,
      "eval_steps_per_second": 0.465,
      "step": 750
    },
    {
      "epoch": 0.2189344799941695,
      "grad_norm": 71.71930402748558,
      "learning_rate": 9.096061004122971e-06,
      "loss": 1.07,
      "step": 751
    },
    {
      "epoch": 0.2192260039355732,
      "grad_norm": 243.72376207675697,
      "learning_rate": 9.09335233817095e-06,
      "loss": 1.2008,
      "step": 752
    },
    {
      "epoch": 0.2195175278769769,
      "grad_norm": 27.723480558135595,
      "learning_rate": 9.09064002457694e-06,
      "loss": 1.0635,
      "step": 753
    },
    {
      "epoch": 0.2198090518183806,
      "grad_norm": 20.424047341101545,
      "learning_rate": 9.08792406575792e-06,
      "loss": 0.9725,
      "step": 754
    },
    {
      "epoch": 0.22010057575978428,
      "grad_norm": 26.325542473643743,
      "learning_rate": 9.085204464134117e-06,
      "loss": 1.0266,
      "step": 755
    },
    {
      "epoch": 0.22039209970118795,
      "grad_norm": 39.47136942273286,
      "learning_rate": 9.082481222129008e-06,
      "loss": 0.9976,
      "step": 756
    },
    {
      "epoch": 0.22068362364259164,
      "grad_norm": 30.062399778935454,
      "learning_rate": 9.07975434216931e-06,
      "loss": 0.9278,
      "step": 757
    },
    {
      "epoch": 0.22097514758399533,
      "grad_norm": 20.64775838320534,
      "learning_rate": 9.07702382668498e-06,
      "loss": 0.9213,
      "step": 758
    },
    {
      "epoch": 0.22126667152539903,
      "grad_norm": 41.95078300817685,
      "learning_rate": 9.074289678109221e-06,
      "loss": 1.1437,
      "step": 759
    },
    {
      "epoch": 0.22155819546680272,
      "grad_norm": 28.598258939797045,
      "learning_rate": 9.071551898878471e-06,
      "loss": 1.1146,
      "step": 760
    },
    {
      "epoch": 0.2218497194082064,
      "grad_norm": 29.20676175380324,
      "learning_rate": 9.068810491432396e-06,
      "loss": 1.0739,
      "step": 761
    },
    {
      "epoch": 0.22214124334961008,
      "grad_norm": 48.3242090009399,
      "learning_rate": 9.066065458213908e-06,
      "loss": 0.9264,
      "step": 762
    },
    {
      "epoch": 0.22243276729101377,
      "grad_norm": 16.044977145885834,
      "learning_rate": 9.063316801669141e-06,
      "loss": 0.8155,
      "step": 763
    },
    {
      "epoch": 0.22272429123241747,
      "grad_norm": 21.962773719454354,
      "learning_rate": 9.06056452424746e-06,
      "loss": 1.0022,
      "step": 764
    },
    {
      "epoch": 0.22301581517382116,
      "grad_norm": 29.165932471680883,
      "learning_rate": 9.057808628401458e-06,
      "loss": 1.1942,
      "step": 765
    },
    {
      "epoch": 0.22330733911522485,
      "grad_norm": 32.57541165365472,
      "learning_rate": 9.055049116586951e-06,
      "loss": 1.0112,
      "step": 766
    },
    {
      "epoch": 0.22359886305662852,
      "grad_norm": 33.80802087601886,
      "learning_rate": 9.052285991262975e-06,
      "loss": 0.9718,
      "step": 767
    },
    {
      "epoch": 0.2238903869980322,
      "grad_norm": 22.24296895760556,
      "learning_rate": 9.049519254891793e-06,
      "loss": 1.0123,
      "step": 768
    },
    {
      "epoch": 0.2241819109394359,
      "grad_norm": 31.957710278044264,
      "learning_rate": 9.046748909938876e-06,
      "loss": 0.8548,
      "step": 769
    },
    {
      "epoch": 0.2244734348808396,
      "grad_norm": 29.854933208354023,
      "learning_rate": 9.04397495887292e-06,
      "loss": 0.906,
      "step": 770
    },
    {
      "epoch": 0.2247649588222433,
      "grad_norm": 46.97918000136825,
      "learning_rate": 9.04119740416583e-06,
      "loss": 0.9782,
      "step": 771
    },
    {
      "epoch": 0.22505648276364695,
      "grad_norm": 47.03829783774376,
      "learning_rate": 9.038416248292725e-06,
      "loss": 0.9607,
      "step": 772
    },
    {
      "epoch": 0.22534800670505065,
      "grad_norm": 20.893458804396165,
      "learning_rate": 9.035631493731927e-06,
      "loss": 1.0099,
      "step": 773
    },
    {
      "epoch": 0.22563953064645434,
      "grad_norm": 42.644115167777855,
      "learning_rate": 9.03284314296497e-06,
      "loss": 1.0342,
      "step": 774
    },
    {
      "epoch": 0.22593105458785803,
      "grad_norm": 21.057092384205763,
      "learning_rate": 9.03005119847659e-06,
      "loss": 1.0133,
      "step": 775
    },
    {
      "epoch": 0.22622257852926173,
      "grad_norm": 37.43385689620047,
      "learning_rate": 9.02725566275473e-06,
      "loss": 1.0482,
      "step": 776
    },
    {
      "epoch": 0.2265141024706654,
      "grad_norm": 23.87570683372111,
      "learning_rate": 9.024456538290528e-06,
      "loss": 1.0429,
      "step": 777
    },
    {
      "epoch": 0.22680562641206908,
      "grad_norm": 30.103135126119362,
      "learning_rate": 9.021653827578322e-06,
      "loss": 0.8901,
      "step": 778
    },
    {
      "epoch": 0.22709715035347278,
      "grad_norm": 32.55356837497784,
      "learning_rate": 9.018847533115645e-06,
      "loss": 1.2135,
      "step": 779
    },
    {
      "epoch": 0.22738867429487647,
      "grad_norm": 37.2727059177145,
      "learning_rate": 9.016037657403225e-06,
      "loss": 0.9164,
      "step": 780
    },
    {
      "epoch": 0.22768019823628016,
      "grad_norm": 22.983829849838678,
      "learning_rate": 9.013224202944982e-06,
      "loss": 1.1413,
      "step": 781
    },
    {
      "epoch": 0.22797172217768386,
      "grad_norm": 23.549880784685012,
      "learning_rate": 9.01040717224802e-06,
      "loss": 0.9373,
      "step": 782
    },
    {
      "epoch": 0.22826324611908752,
      "grad_norm": 62.412859743543365,
      "learning_rate": 9.007586567822634e-06,
      "loss": 0.9632,
      "step": 783
    },
    {
      "epoch": 0.2285547700604912,
      "grad_norm": 18.10773663331525,
      "learning_rate": 9.004762392182307e-06,
      "loss": 0.9277,
      "step": 784
    },
    {
      "epoch": 0.2288462940018949,
      "grad_norm": 58.84786121639291,
      "learning_rate": 9.001934647843697e-06,
      "loss": 1.0521,
      "step": 785
    },
    {
      "epoch": 0.2291378179432986,
      "grad_norm": 21.974297336067234,
      "learning_rate": 8.999103337326646e-06,
      "loss": 1.0252,
      "step": 786
    },
    {
      "epoch": 0.2294293418847023,
      "grad_norm": 19.55555603964135,
      "learning_rate": 8.996268463154175e-06,
      "loss": 1.05,
      "step": 787
    },
    {
      "epoch": 0.22972086582610596,
      "grad_norm": 43.31531959385443,
      "learning_rate": 8.993430027852476e-06,
      "loss": 0.9535,
      "step": 788
    },
    {
      "epoch": 0.23001238976750965,
      "grad_norm": 53.42220378481584,
      "learning_rate": 8.99058803395092e-06,
      "loss": 1.0742,
      "step": 789
    },
    {
      "epoch": 0.23030391370891334,
      "grad_norm": 61.27425816136208,
      "learning_rate": 8.987742483982044e-06,
      "loss": 1.18,
      "step": 790
    },
    {
      "epoch": 0.23059543765031704,
      "grad_norm": 28.992086907546188,
      "learning_rate": 8.984893380481559e-06,
      "loss": 1.1363,
      "step": 791
    },
    {
      "epoch": 0.23088696159172073,
      "grad_norm": 29.915367279135406,
      "learning_rate": 8.982040725988337e-06,
      "loss": 0.9132,
      "step": 792
    },
    {
      "epoch": 0.2311784855331244,
      "grad_norm": 25.45694740490939,
      "learning_rate": 8.979184523044419e-06,
      "loss": 0.9308,
      "step": 793
    },
    {
      "epoch": 0.2314700094745281,
      "grad_norm": 234.1297444769745,
      "learning_rate": 8.976324774195005e-06,
      "loss": 1.04,
      "step": 794
    },
    {
      "epoch": 0.23176153341593178,
      "grad_norm": 32.63631036899571,
      "learning_rate": 8.973461481988459e-06,
      "loss": 0.938,
      "step": 795
    },
    {
      "epoch": 0.23205305735733547,
      "grad_norm": 39.46922189115229,
      "learning_rate": 8.970594648976299e-06,
      "loss": 0.9156,
      "step": 796
    },
    {
      "epoch": 0.23234458129873917,
      "grad_norm": 25.487723098260382,
      "learning_rate": 8.967724277713194e-06,
      "loss": 1.0144,
      "step": 797
    },
    {
      "epoch": 0.23263610524014286,
      "grad_norm": 54.031193870023785,
      "learning_rate": 8.964850370756978e-06,
      "loss": 1.0944,
      "step": 798
    },
    {
      "epoch": 0.23292762918154652,
      "grad_norm": 28.253251412935338,
      "learning_rate": 8.961972930668625e-06,
      "loss": 1.0066,
      "step": 799
    },
    {
      "epoch": 0.23321915312295022,
      "grad_norm": 18.53170050447928,
      "learning_rate": 8.95909196001226e-06,
      "loss": 1.0573,
      "step": 800
    },
    {
      "epoch": 0.23321915312295022,
      "eval_loss": 1.1618326902389526,
      "eval_runtime": 2103.3869,
      "eval_samples_per_second": 8.211,
      "eval_steps_per_second": 0.456,
      "step": 800
    },
    {
      "epoch": 0.2335106770643539,
      "grad_norm": 40.27261175268867,
      "learning_rate": 8.956207461355158e-06,
      "loss": 1.1199,
      "step": 801
    },
    {
      "epoch": 0.2338022010057576,
      "grad_norm": 17.866542418804233,
      "learning_rate": 8.953319437267731e-06,
      "loss": 0.7901,
      "step": 802
    },
    {
      "epoch": 0.2340937249471613,
      "grad_norm": 20.51964657421649,
      "learning_rate": 8.950427890323541e-06,
      "loss": 0.9383,
      "step": 803
    },
    {
      "epoch": 0.23438524888856496,
      "grad_norm": 23.22793624806918,
      "learning_rate": 8.947532823099284e-06,
      "loss": 1.0735,
      "step": 804
    },
    {
      "epoch": 0.23467677282996866,
      "grad_norm": 93.55215793425462,
      "learning_rate": 8.944634238174791e-06,
      "loss": 1.067,
      "step": 805
    },
    {
      "epoch": 0.23496829677137235,
      "grad_norm": 23.819901880997076,
      "learning_rate": 8.941732138133032e-06,
      "loss": 1.0248,
      "step": 806
    },
    {
      "epoch": 0.23525982071277604,
      "grad_norm": 67.9703273717819,
      "learning_rate": 8.93882652556011e-06,
      "loss": 1.0198,
      "step": 807
    },
    {
      "epoch": 0.23555134465417973,
      "grad_norm": 19.48474708574522,
      "learning_rate": 8.935917403045251e-06,
      "loss": 0.9286,
      "step": 808
    },
    {
      "epoch": 0.2358428685955834,
      "grad_norm": 66.20845972074659,
      "learning_rate": 8.933004773180819e-06,
      "loss": 1.0763,
      "step": 809
    },
    {
      "epoch": 0.2361343925369871,
      "grad_norm": 48.73672734229711,
      "learning_rate": 8.930088638562296e-06,
      "loss": 1.13,
      "step": 810
    },
    {
      "epoch": 0.23642591647839079,
      "grad_norm": 38.03156736748355,
      "learning_rate": 8.927169001788286e-06,
      "loss": 0.9738,
      "step": 811
    },
    {
      "epoch": 0.23671744041979448,
      "grad_norm": 28.618707773838835,
      "learning_rate": 8.924245865460523e-06,
      "loss": 0.9813,
      "step": 812
    },
    {
      "epoch": 0.23700896436119817,
      "grad_norm": 178.15957500173664,
      "learning_rate": 8.921319232183851e-06,
      "loss": 1.1155,
      "step": 813
    },
    {
      "epoch": 0.23730048830260186,
      "grad_norm": 18.59654704086866,
      "learning_rate": 8.918389104566232e-06,
      "loss": 0.9212,
      "step": 814
    },
    {
      "epoch": 0.23759201224400553,
      "grad_norm": 36.61535960093206,
      "learning_rate": 8.915455485218743e-06,
      "loss": 1.0332,
      "step": 815
    },
    {
      "epoch": 0.23788353618540922,
      "grad_norm": 75.09126353807024,
      "learning_rate": 8.912518376755572e-06,
      "loss": 1.1163,
      "step": 816
    },
    {
      "epoch": 0.23817506012681292,
      "grad_norm": 23.72647437120418,
      "learning_rate": 8.909577781794016e-06,
      "loss": 1.1174,
      "step": 817
    },
    {
      "epoch": 0.2384665840682166,
      "grad_norm": 23.68866834946315,
      "learning_rate": 8.906633702954482e-06,
      "loss": 1.0231,
      "step": 818
    },
    {
      "epoch": 0.2387581080096203,
      "grad_norm": 38.325144198843645,
      "learning_rate": 8.903686142860473e-06,
      "loss": 0.905,
      "step": 819
    },
    {
      "epoch": 0.23904963195102397,
      "grad_norm": 48.35500076239675,
      "learning_rate": 8.900735104138605e-06,
      "loss": 0.78,
      "step": 820
    },
    {
      "epoch": 0.23934115589242766,
      "grad_norm": 17.155291745742403,
      "learning_rate": 8.897780589418588e-06,
      "loss": 1.0765,
      "step": 821
    },
    {
      "epoch": 0.23963267983383135,
      "grad_norm": 27.011117145089475,
      "learning_rate": 8.894822601333228e-06,
      "loss": 0.8654,
      "step": 822
    },
    {
      "epoch": 0.23992420377523505,
      "grad_norm": 22.556476737179707,
      "learning_rate": 8.89186114251843e-06,
      "loss": 1.0477,
      "step": 823
    },
    {
      "epoch": 0.24021572771663874,
      "grad_norm": 15.306669839035273,
      "learning_rate": 8.888896215613192e-06,
      "loss": 0.9563,
      "step": 824
    },
    {
      "epoch": 0.2405072516580424,
      "grad_norm": 42.365152173892064,
      "learning_rate": 8.885927823259597e-06,
      "loss": 1.0081,
      "step": 825
    },
    {
      "epoch": 0.2407987755994461,
      "grad_norm": 36.45258254100991,
      "learning_rate": 8.882955968102822e-06,
      "loss": 1.0676,
      "step": 826
    },
    {
      "epoch": 0.2410902995408498,
      "grad_norm": 18.34110636383369,
      "learning_rate": 8.879980652791126e-06,
      "loss": 0.8663,
      "step": 827
    },
    {
      "epoch": 0.24138182348225348,
      "grad_norm": 20.366025776836047,
      "learning_rate": 8.877001879975857e-06,
      "loss": 1.085,
      "step": 828
    },
    {
      "epoch": 0.24167334742365718,
      "grad_norm": 26.938668891732785,
      "learning_rate": 8.874019652311437e-06,
      "loss": 0.9265,
      "step": 829
    },
    {
      "epoch": 0.24196487136506087,
      "grad_norm": 109.39735135693614,
      "learning_rate": 8.87103397245537e-06,
      "loss": 0.9927,
      "step": 830
    },
    {
      "epoch": 0.24225639530646453,
      "grad_norm": 36.91376634833503,
      "learning_rate": 8.868044843068234e-06,
      "loss": 0.9981,
      "step": 831
    },
    {
      "epoch": 0.24254791924786823,
      "grad_norm": 17.00519222191051,
      "learning_rate": 8.865052266813686e-06,
      "loss": 1.2373,
      "step": 832
    },
    {
      "epoch": 0.24283944318927192,
      "grad_norm": 65.01749362399237,
      "learning_rate": 8.86205624635845e-06,
      "loss": 1.0412,
      "step": 833
    },
    {
      "epoch": 0.2431309671306756,
      "grad_norm": 30.518587393875176,
      "learning_rate": 8.85905678437232e-06,
      "loss": 1.271,
      "step": 834
    },
    {
      "epoch": 0.2434224910720793,
      "grad_norm": 41.48430784365996,
      "learning_rate": 8.85605388352816e-06,
      "loss": 0.9506,
      "step": 835
    },
    {
      "epoch": 0.24371401501348297,
      "grad_norm": 27.735651464723688,
      "learning_rate": 8.853047546501893e-06,
      "loss": 1.0423,
      "step": 836
    },
    {
      "epoch": 0.24400553895488666,
      "grad_norm": 16.21268504361212,
      "learning_rate": 8.850037775972508e-06,
      "loss": 0.9782,
      "step": 837
    },
    {
      "epoch": 0.24429706289629036,
      "grad_norm": 19.692981601313118,
      "learning_rate": 8.847024574622051e-06,
      "loss": 1.0961,
      "step": 838
    },
    {
      "epoch": 0.24458858683769405,
      "grad_norm": 16.855174425137488,
      "learning_rate": 8.84400794513563e-06,
      "loss": 0.9096,
      "step": 839
    },
    {
      "epoch": 0.24488011077909774,
      "grad_norm": 22.2016414043827,
      "learning_rate": 8.840987890201404e-06,
      "loss": 1.0723,
      "step": 840
    },
    {
      "epoch": 0.2451716347205014,
      "grad_norm": 13.893798450147065,
      "learning_rate": 8.83796441251058e-06,
      "loss": 0.9919,
      "step": 841
    },
    {
      "epoch": 0.2454631586619051,
      "grad_norm": 47.610534657618494,
      "learning_rate": 8.834937514757428e-06,
      "loss": 0.898,
      "step": 842
    },
    {
      "epoch": 0.2457546826033088,
      "grad_norm": 26.86996358883563,
      "learning_rate": 8.831907199639252e-06,
      "loss": 1.0464,
      "step": 843
    },
    {
      "epoch": 0.2460462065447125,
      "grad_norm": 19.378126113364743,
      "learning_rate": 8.828873469856408e-06,
      "loss": 1.1723,
      "step": 844
    },
    {
      "epoch": 0.24633773048611618,
      "grad_norm": 21.767868153792353,
      "learning_rate": 8.825836328112296e-06,
      "loss": 1.0738,
      "step": 845
    },
    {
      "epoch": 0.24662925442751987,
      "grad_norm": 45.16156352044694,
      "learning_rate": 8.822795777113352e-06,
      "loss": 1.0139,
      "step": 846
    },
    {
      "epoch": 0.24692077836892354,
      "grad_norm": 15.779908501210965,
      "learning_rate": 8.819751819569053e-06,
      "loss": 0.8201,
      "step": 847
    },
    {
      "epoch": 0.24721230231032723,
      "grad_norm": 36.390515464247414,
      "learning_rate": 8.816704458191913e-06,
      "loss": 1.0434,
      "step": 848
    },
    {
      "epoch": 0.24750382625173092,
      "grad_norm": 26.86066724095118,
      "learning_rate": 8.813653695697475e-06,
      "loss": 0.9588,
      "step": 849
    },
    {
      "epoch": 0.24779535019313462,
      "grad_norm": 23.41446707947497,
      "learning_rate": 8.810599534804315e-06,
      "loss": 0.9988,
      "step": 850
    },
    {
      "epoch": 0.24779535019313462,
      "eval_loss": 1.1622024774551392,
      "eval_runtime": 2043.4216,
      "eval_samples_per_second": 8.452,
      "eval_steps_per_second": 0.47,
      "step": 850
    },
    {
      "epoch": 0.2480868741345383,
      "grad_norm": 45.70152583629363,
      "learning_rate": 8.807541978234038e-06,
      "loss": 0.9119,
      "step": 851
    },
    {
      "epoch": 0.24837839807594198,
      "grad_norm": 33.98667436393294,
      "learning_rate": 8.804481028711274e-06,
      "loss": 1.0322,
      "step": 852
    },
    {
      "epoch": 0.24866992201734567,
      "grad_norm": 26.86494236658137,
      "learning_rate": 8.80141668896368e-06,
      "loss": 0.9967,
      "step": 853
    },
    {
      "epoch": 0.24896144595874936,
      "grad_norm": 26.30357174025587,
      "learning_rate": 8.798348961721925e-06,
      "loss": 1.0771,
      "step": 854
    },
    {
      "epoch": 0.24925296990015305,
      "grad_norm": 35.42110258283485,
      "learning_rate": 8.795277849719706e-06,
      "loss": 0.8442,
      "step": 855
    },
    {
      "epoch": 0.24954449384155675,
      "grad_norm": 17.763180366270735,
      "learning_rate": 8.792203355693731e-06,
      "loss": 0.8855,
      "step": 856
    },
    {
      "epoch": 0.2498360177829604,
      "grad_norm": 24.231485523004952,
      "learning_rate": 8.789125482383726e-06,
      "loss": 0.9663,
      "step": 857
    },
    {
      "epoch": 0.25012754172436413,
      "grad_norm": 39.70650011758068,
      "learning_rate": 8.786044232532423e-06,
      "loss": 1.0081,
      "step": 858
    },
    {
      "epoch": 0.2504190656657678,
      "grad_norm": 36.82294640930288,
      "learning_rate": 8.782959608885568e-06,
      "loss": 0.9242,
      "step": 859
    },
    {
      "epoch": 0.25071058960717146,
      "grad_norm": 20.649323537475457,
      "learning_rate": 8.77987161419191e-06,
      "loss": 1.038,
      "step": 860
    },
    {
      "epoch": 0.25100211354857516,
      "grad_norm": 56.537040198408604,
      "learning_rate": 8.776780251203205e-06,
      "loss": 1.0151,
      "step": 861
    },
    {
      "epoch": 0.25129363748997885,
      "grad_norm": 24.62965439297132,
      "learning_rate": 8.773685522674205e-06,
      "loss": 1.0969,
      "step": 862
    },
    {
      "epoch": 0.25158516143138254,
      "grad_norm": 16.347622480502448,
      "learning_rate": 8.770587431362667e-06,
      "loss": 0.9512,
      "step": 863
    },
    {
      "epoch": 0.25187668537278624,
      "grad_norm": 74.00475591847797,
      "learning_rate": 8.767485980029342e-06,
      "loss": 0.963,
      "step": 864
    },
    {
      "epoch": 0.25216820931418993,
      "grad_norm": 14.6627631661899,
      "learning_rate": 8.764381171437976e-06,
      "loss": 0.9047,
      "step": 865
    },
    {
      "epoch": 0.2524597332555936,
      "grad_norm": 58.37394463100665,
      "learning_rate": 8.761273008355306e-06,
      "loss": 0.9486,
      "step": 866
    },
    {
      "epoch": 0.2527512571969973,
      "grad_norm": 22.99035829849301,
      "learning_rate": 8.758161493551055e-06,
      "loss": 0.8625,
      "step": 867
    },
    {
      "epoch": 0.253042781138401,
      "grad_norm": 27.596765687937232,
      "learning_rate": 8.755046629797944e-06,
      "loss": 0.9958,
      "step": 868
    },
    {
      "epoch": 0.2533343050798047,
      "grad_norm": 18.792551390197094,
      "learning_rate": 8.751928419871662e-06,
      "loss": 0.9755,
      "step": 869
    },
    {
      "epoch": 0.2536258290212084,
      "grad_norm": 20.826937690356452,
      "learning_rate": 8.748806866550895e-06,
      "loss": 1.0692,
      "step": 870
    },
    {
      "epoch": 0.25391735296261203,
      "grad_norm": 24.207784077583746,
      "learning_rate": 8.745681972617298e-06,
      "loss": 1.1657,
      "step": 871
    },
    {
      "epoch": 0.2542088769040157,
      "grad_norm": 25.866543195947575,
      "learning_rate": 8.742553740855507e-06,
      "loss": 1.0625,
      "step": 872
    },
    {
      "epoch": 0.2545004008454194,
      "grad_norm": 73.07684596249292,
      "learning_rate": 8.739422174053131e-06,
      "loss": 1.0716,
      "step": 873
    },
    {
      "epoch": 0.2547919247868231,
      "grad_norm": 14.522911274657778,
      "learning_rate": 8.736287275000755e-06,
      "loss": 0.949,
      "step": 874
    },
    {
      "epoch": 0.2550834487282268,
      "grad_norm": 24.54841640989312,
      "learning_rate": 8.733149046491926e-06,
      "loss": 1.1301,
      "step": 875
    },
    {
      "epoch": 0.2553749726696305,
      "grad_norm": 21.46685443357537,
      "learning_rate": 8.730007491323167e-06,
      "loss": 1.0942,
      "step": 876
    },
    {
      "epoch": 0.2556664966110342,
      "grad_norm": 12.071138646354624,
      "learning_rate": 8.726862612293957e-06,
      "loss": 0.9022,
      "step": 877
    },
    {
      "epoch": 0.2559580205524379,
      "grad_norm": 43.05900704995226,
      "learning_rate": 8.723714412206741e-06,
      "loss": 0.9036,
      "step": 878
    },
    {
      "epoch": 0.2562495444938416,
      "grad_norm": 16.2986531318287,
      "learning_rate": 8.720562893866924e-06,
      "loss": 0.9732,
      "step": 879
    },
    {
      "epoch": 0.25654106843524527,
      "grad_norm": 29.40026991467079,
      "learning_rate": 8.717408060082865e-06,
      "loss": 0.9625,
      "step": 880
    },
    {
      "epoch": 0.2568325923766489,
      "grad_norm": 51.820203077274,
      "learning_rate": 8.714249913665883e-06,
      "loss": 0.8774,
      "step": 881
    },
    {
      "epoch": 0.2571241163180526,
      "grad_norm": 20.25707264815185,
      "learning_rate": 8.711088457430239e-06,
      "loss": 1.1145,
      "step": 882
    },
    {
      "epoch": 0.2574156402594563,
      "grad_norm": 25.35784647534397,
      "learning_rate": 8.707923694193156e-06,
      "loss": 1.1178,
      "step": 883
    },
    {
      "epoch": 0.25770716420086,
      "grad_norm": 18.325052530228934,
      "learning_rate": 8.704755626774796e-06,
      "loss": 0.8555,
      "step": 884
    },
    {
      "epoch": 0.2579986881422637,
      "grad_norm": 18.691396139684954,
      "learning_rate": 8.701584257998266e-06,
      "loss": 1.0115,
      "step": 885
    },
    {
      "epoch": 0.25829021208366737,
      "grad_norm": 21.441808421578596,
      "learning_rate": 8.698409590689616e-06,
      "loss": 0.9131,
      "step": 886
    },
    {
      "epoch": 0.25858173602507106,
      "grad_norm": 61.22344888726197,
      "learning_rate": 8.695231627677834e-06,
      "loss": 1.057,
      "step": 887
    },
    {
      "epoch": 0.25887325996647476,
      "grad_norm": 16.91331244211354,
      "learning_rate": 8.692050371794849e-06,
      "loss": 1.0739,
      "step": 888
    },
    {
      "epoch": 0.25916478390787845,
      "grad_norm": 49.56608745942036,
      "learning_rate": 8.68886582587552e-06,
      "loss": 1.0097,
      "step": 889
    },
    {
      "epoch": 0.25945630784928214,
      "grad_norm": 17.84339669136109,
      "learning_rate": 8.685677992757637e-06,
      "loss": 1.1252,
      "step": 890
    },
    {
      "epoch": 0.25974783179068583,
      "grad_norm": 26.336171053162037,
      "learning_rate": 8.682486875281925e-06,
      "loss": 0.8771,
      "step": 891
    },
    {
      "epoch": 0.26003935573208947,
      "grad_norm": 39.771173523409104,
      "learning_rate": 8.67929247629203e-06,
      "loss": 0.9752,
      "step": 892
    },
    {
      "epoch": 0.26033087967349317,
      "grad_norm": 19.25607707586375,
      "learning_rate": 8.676094798634527e-06,
      "loss": 0.9137,
      "step": 893
    },
    {
      "epoch": 0.26062240361489686,
      "grad_norm": 37.284584749434956,
      "learning_rate": 8.672893845158908e-06,
      "loss": 1.1288,
      "step": 894
    },
    {
      "epoch": 0.26091392755630055,
      "grad_norm": 36.168529033193295,
      "learning_rate": 8.669689618717584e-06,
      "loss": 0.9637,
      "step": 895
    },
    {
      "epoch": 0.26120545149770424,
      "grad_norm": 17.26094904931096,
      "learning_rate": 8.66648212216589e-06,
      "loss": 1.0841,
      "step": 896
    },
    {
      "epoch": 0.26149697543910794,
      "grad_norm": 32.577290915533666,
      "learning_rate": 8.663271358362064e-06,
      "loss": 1.0545,
      "step": 897
    },
    {
      "epoch": 0.26178849938051163,
      "grad_norm": 20.7087249535045,
      "learning_rate": 8.660057330167267e-06,
      "loss": 1.1164,
      "step": 898
    },
    {
      "epoch": 0.2620800233219153,
      "grad_norm": 96.67116666317605,
      "learning_rate": 8.656840040445562e-06,
      "loss": 1.0883,
      "step": 899
    },
    {
      "epoch": 0.262371547263319,
      "grad_norm": 32.40128672801176,
      "learning_rate": 8.653619492063916e-06,
      "loss": 1.0185,
      "step": 900
    },
    {
      "epoch": 0.262371547263319,
      "eval_loss": 1.151246428489685,
      "eval_runtime": 2079.7016,
      "eval_samples_per_second": 8.304,
      "eval_steps_per_second": 0.462,
      "step": 900
    },
    {
      "epoch": 0.2626630712047227,
      "grad_norm": 18.302575831613954,
      "learning_rate": 8.650395687892206e-06,
      "loss": 1.1228,
      "step": 901
    },
    {
      "epoch": 0.2629545951461264,
      "grad_norm": 17.050038627814516,
      "learning_rate": 8.647168630803208e-06,
      "loss": 0.997,
      "step": 902
    },
    {
      "epoch": 0.26324611908753004,
      "grad_norm": 30.79201604727001,
      "learning_rate": 8.643938323672598e-06,
      "loss": 0.9641,
      "step": 903
    },
    {
      "epoch": 0.26353764302893373,
      "grad_norm": 24.248187610477455,
      "learning_rate": 8.640704769378943e-06,
      "loss": 1.0943,
      "step": 904
    },
    {
      "epoch": 0.2638291669703374,
      "grad_norm": 12.903195102597977,
      "learning_rate": 8.63746797080371e-06,
      "loss": 0.8267,
      "step": 905
    },
    {
      "epoch": 0.2641206909117411,
      "grad_norm": 37.89897821366868,
      "learning_rate": 8.634227930831252e-06,
      "loss": 1.0224,
      "step": 906
    },
    {
      "epoch": 0.2644122148531448,
      "grad_norm": 33.35694361146605,
      "learning_rate": 8.630984652348815e-06,
      "loss": 1.0045,
      "step": 907
    },
    {
      "epoch": 0.2647037387945485,
      "grad_norm": 21.750856428085548,
      "learning_rate": 8.627738138246529e-06,
      "loss": 1.1087,
      "step": 908
    },
    {
      "epoch": 0.2649952627359522,
      "grad_norm": 48.075551787112595,
      "learning_rate": 8.624488391417404e-06,
      "loss": 1.1456,
      "step": 909
    },
    {
      "epoch": 0.2652867866773559,
      "grad_norm": 19.440787447204375,
      "learning_rate": 8.621235414757337e-06,
      "loss": 0.9606,
      "step": 910
    },
    {
      "epoch": 0.2655783106187596,
      "grad_norm": 19.373254903059795,
      "learning_rate": 8.617979211165096e-06,
      "loss": 0.9718,
      "step": 911
    },
    {
      "epoch": 0.2658698345601633,
      "grad_norm": 15.705220575062754,
      "learning_rate": 8.61471978354233e-06,
      "loss": 1.0051,
      "step": 912
    },
    {
      "epoch": 0.2661613585015669,
      "grad_norm": 22.941200697983128,
      "learning_rate": 8.611457134793562e-06,
      "loss": 0.9945,
      "step": 913
    },
    {
      "epoch": 0.2664528824429706,
      "grad_norm": 17.647390234731844,
      "learning_rate": 8.608191267826179e-06,
      "loss": 0.979,
      "step": 914
    },
    {
      "epoch": 0.2667444063843743,
      "grad_norm": 18.03594591540412,
      "learning_rate": 8.60492218555044e-06,
      "loss": 0.7965,
      "step": 915
    },
    {
      "epoch": 0.267035930325778,
      "grad_norm": 60.068909666272475,
      "learning_rate": 8.60164989087947e-06,
      "loss": 1.0996,
      "step": 916
    },
    {
      "epoch": 0.2673274542671817,
      "grad_norm": 18.130326714165374,
      "learning_rate": 8.598374386729258e-06,
      "loss": 1.0317,
      "step": 917
    },
    {
      "epoch": 0.2676189782085854,
      "grad_norm": 13.247140787873807,
      "learning_rate": 8.595095676018645e-06,
      "loss": 0.9084,
      "step": 918
    },
    {
      "epoch": 0.26791050214998907,
      "grad_norm": 28.081190368835358,
      "learning_rate": 8.591813761669336e-06,
      "loss": 1.0096,
      "step": 919
    },
    {
      "epoch": 0.26820202609139276,
      "grad_norm": 71.62697650785337,
      "learning_rate": 8.588528646605893e-06,
      "loss": 0.8417,
      "step": 920
    },
    {
      "epoch": 0.26849355003279646,
      "grad_norm": 20.29115522122536,
      "learning_rate": 8.585240333755723e-06,
      "loss": 1.1504,
      "step": 921
    },
    {
      "epoch": 0.26878507397420015,
      "grad_norm": 21.771033503717636,
      "learning_rate": 8.581948826049086e-06,
      "loss": 1.0645,
      "step": 922
    },
    {
      "epoch": 0.26907659791560384,
      "grad_norm": 20.76415837154338,
      "learning_rate": 8.578654126419094e-06,
      "loss": 1.1513,
      "step": 923
    },
    {
      "epoch": 0.2693681218570075,
      "grad_norm": 35.30161604863573,
      "learning_rate": 8.575356237801695e-06,
      "loss": 1.0935,
      "step": 924
    },
    {
      "epoch": 0.2696596457984112,
      "grad_norm": 24.553610952177003,
      "learning_rate": 8.572055163135682e-06,
      "loss": 0.9139,
      "step": 925
    },
    {
      "epoch": 0.26995116973981487,
      "grad_norm": 21.858421670113433,
      "learning_rate": 8.56875090536269e-06,
      "loss": 1.1687,
      "step": 926
    },
    {
      "epoch": 0.27024269368121856,
      "grad_norm": 17.851471969389863,
      "learning_rate": 8.565443467427185e-06,
      "loss": 0.8676,
      "step": 927
    },
    {
      "epoch": 0.27053421762262225,
      "grad_norm": 20.18503988728097,
      "learning_rate": 8.562132852276474e-06,
      "loss": 1.0201,
      "step": 928
    },
    {
      "epoch": 0.27082574156402595,
      "grad_norm": 27.959190698096492,
      "learning_rate": 8.558819062860689e-06,
      "loss": 1.1292,
      "step": 929
    },
    {
      "epoch": 0.27111726550542964,
      "grad_norm": 34.550291900983375,
      "learning_rate": 8.555502102132792e-06,
      "loss": 0.9613,
      "step": 930
    },
    {
      "epoch": 0.27140878944683333,
      "grad_norm": 28.046749553385155,
      "learning_rate": 8.552181973048572e-06,
      "loss": 1.0531,
      "step": 931
    },
    {
      "epoch": 0.271700313388237,
      "grad_norm": 35.65758802956286,
      "learning_rate": 8.548858678566643e-06,
      "loss": 0.9819,
      "step": 932
    },
    {
      "epoch": 0.2719918373296407,
      "grad_norm": 20.136880877349032,
      "learning_rate": 8.545532221648436e-06,
      "loss": 1.0772,
      "step": 933
    },
    {
      "epoch": 0.2722833612710444,
      "grad_norm": 46.25130408996463,
      "learning_rate": 8.542202605258204e-06,
      "loss": 0.8607,
      "step": 934
    },
    {
      "epoch": 0.27257488521244805,
      "grad_norm": 18.00137256990522,
      "learning_rate": 8.53886983236301e-06,
      "loss": 1.1376,
      "step": 935
    },
    {
      "epoch": 0.27286640915385174,
      "grad_norm": 19.207451245547823,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.9328,
      "step": 936
    },
    {
      "epoch": 0.27315793309525543,
      "grad_norm": 20.45854909732303,
      "learning_rate": 8.532194828940074e-06,
      "loss": 1.0734,
      "step": 937
    },
    {
      "epoch": 0.2734494570366591,
      "grad_norm": 42.70969477517702,
      "learning_rate": 8.528852604360518e-06,
      "loss": 0.9913,
      "step": 938
    },
    {
      "epoch": 0.2737409809780628,
      "grad_norm": 28.45857513228677,
      "learning_rate": 8.52550723517237e-06,
      "loss": 0.9756,
      "step": 939
    },
    {
      "epoch": 0.2740325049194665,
      "grad_norm": 19.596479657450534,
      "learning_rate": 8.52215872435673e-06,
      "loss": 1.0035,
      "step": 940
    },
    {
      "epoch": 0.2743240288608702,
      "grad_norm": 17.45701328647761,
      "learning_rate": 8.518807074897509e-06,
      "loss": 0.9126,
      "step": 941
    },
    {
      "epoch": 0.2746155528022739,
      "grad_norm": 18.593165655588713,
      "learning_rate": 8.515452289781403e-06,
      "loss": 0.8931,
      "step": 942
    },
    {
      "epoch": 0.2749070767436776,
      "grad_norm": 21.996067104819904,
      "learning_rate": 8.512094371997908e-06,
      "loss": 0.8753,
      "step": 943
    },
    {
      "epoch": 0.2751986006850813,
      "grad_norm": 28.687970072967424,
      "learning_rate": 8.50873332453931e-06,
      "loss": 0.9016,
      "step": 944
    },
    {
      "epoch": 0.2754901246264849,
      "grad_norm": 16.320492097914922,
      "learning_rate": 8.505369150400685e-06,
      "loss": 0.846,
      "step": 945
    },
    {
      "epoch": 0.2757816485678886,
      "grad_norm": 23.81868956044284,
      "learning_rate": 8.50200185257989e-06,
      "loss": 0.924,
      "step": 946
    },
    {
      "epoch": 0.2760731725092923,
      "grad_norm": 28.53145285875362,
      "learning_rate": 8.498631434077577e-06,
      "loss": 1.2219,
      "step": 947
    },
    {
      "epoch": 0.276364696450696,
      "grad_norm": 54.03999746600719,
      "learning_rate": 8.495257897897166e-06,
      "loss": 1.0314,
      "step": 948
    },
    {
      "epoch": 0.2766562203920997,
      "grad_norm": 35.62738153455295,
      "learning_rate": 8.491881247044866e-06,
      "loss": 0.9784,
      "step": 949
    },
    {
      "epoch": 0.2769477443335034,
      "grad_norm": 34.14273763829868,
      "learning_rate": 8.48850148452965e-06,
      "loss": 1.0292,
      "step": 950
    },
    {
      "epoch": 0.2769477443335034,
      "eval_loss": 1.1573375463485718,
      "eval_runtime": 2051.9871,
      "eval_samples_per_second": 8.416,
      "eval_steps_per_second": 0.468,
      "step": 950
    },
    {
      "epoch": 0.2772392682749071,
      "grad_norm": 37.88817132434681,
      "learning_rate": 8.485118613363276e-06,
      "loss": 0.9638,
      "step": 951
    },
    {
      "epoch": 0.2775307922163108,
      "grad_norm": 15.973194609739432,
      "learning_rate": 8.481732636560266e-06,
      "loss": 1.0605,
      "step": 952
    },
    {
      "epoch": 0.27782231615771447,
      "grad_norm": 17.444964809749628,
      "learning_rate": 8.478343557137907e-06,
      "loss": 0.9648,
      "step": 953
    },
    {
      "epoch": 0.27811384009911816,
      "grad_norm": 27.399230362935306,
      "learning_rate": 8.474951378116253e-06,
      "loss": 1.0887,
      "step": 954
    },
    {
      "epoch": 0.27840536404052185,
      "grad_norm": 46.24196349593448,
      "learning_rate": 8.471556102518129e-06,
      "loss": 1.1864,
      "step": 955
    },
    {
      "epoch": 0.2786968879819255,
      "grad_norm": 22.75259518859823,
      "learning_rate": 8.468157733369102e-06,
      "loss": 1.1722,
      "step": 956
    },
    {
      "epoch": 0.2789884119233292,
      "grad_norm": 26.740322045670123,
      "learning_rate": 8.46475627369751e-06,
      "loss": 0.9037,
      "step": 957
    },
    {
      "epoch": 0.2792799358647329,
      "grad_norm": 68.54682373594649,
      "learning_rate": 8.461351726534438e-06,
      "loss": 1.0804,
      "step": 958
    },
    {
      "epoch": 0.27957145980613657,
      "grad_norm": 25.278180356932673,
      "learning_rate": 8.457944094913724e-06,
      "loss": 1.063,
      "step": 959
    },
    {
      "epoch": 0.27986298374754026,
      "grad_norm": 42.1744597677913,
      "learning_rate": 8.454533381871957e-06,
      "loss": 1.0517,
      "step": 960
    },
    {
      "epoch": 0.28015450768894395,
      "grad_norm": 17.15532402676699,
      "learning_rate": 8.45111959044847e-06,
      "loss": 0.9836,
      "step": 961
    },
    {
      "epoch": 0.28044603163034765,
      "grad_norm": 20.069327478838023,
      "learning_rate": 8.447702723685335e-06,
      "loss": 0.983,
      "step": 962
    },
    {
      "epoch": 0.28073755557175134,
      "grad_norm": 29.168306339463605,
      "learning_rate": 8.444282784627374e-06,
      "loss": 0.8921,
      "step": 963
    },
    {
      "epoch": 0.28102907951315503,
      "grad_norm": 32.1212173799936,
      "learning_rate": 8.440859776322137e-06,
      "loss": 0.9716,
      "step": 964
    },
    {
      "epoch": 0.2813206034545587,
      "grad_norm": 15.818289257957133,
      "learning_rate": 8.437433701819914e-06,
      "loss": 1.0033,
      "step": 965
    },
    {
      "epoch": 0.2816121273959624,
      "grad_norm": 32.777811217547764,
      "learning_rate": 8.43400456417373e-06,
      "loss": 0.9358,
      "step": 966
    },
    {
      "epoch": 0.28190365133736606,
      "grad_norm": 26.02526243583659,
      "learning_rate": 8.430572366439332e-06,
      "loss": 0.9551,
      "step": 967
    },
    {
      "epoch": 0.28219517527876975,
      "grad_norm": 15.455017815962172,
      "learning_rate": 8.4271371116752e-06,
      "loss": 0.9738,
      "step": 968
    },
    {
      "epoch": 0.28248669922017344,
      "grad_norm": 21.846152768574175,
      "learning_rate": 8.423698802942536e-06,
      "loss": 0.9987,
      "step": 969
    },
    {
      "epoch": 0.28277822316157714,
      "grad_norm": 23.52205073894214,
      "learning_rate": 8.420257443305264e-06,
      "loss": 0.9462,
      "step": 970
    },
    {
      "epoch": 0.28306974710298083,
      "grad_norm": 24.273341480809453,
      "learning_rate": 8.416813035830025e-06,
      "loss": 0.962,
      "step": 971
    },
    {
      "epoch": 0.2833612710443845,
      "grad_norm": 19.99964360315538,
      "learning_rate": 8.41336558358618e-06,
      "loss": 1.0153,
      "step": 972
    },
    {
      "epoch": 0.2836527949857882,
      "grad_norm": 30.80863300221459,
      "learning_rate": 8.4099150896458e-06,
      "loss": 1.0894,
      "step": 973
    },
    {
      "epoch": 0.2839443189271919,
      "grad_norm": 30.01147600207258,
      "learning_rate": 8.406461557083666e-06,
      "loss": 0.9685,
      "step": 974
    },
    {
      "epoch": 0.2842358428685956,
      "grad_norm": 31.99525877858003,
      "learning_rate": 8.403004988977267e-06,
      "loss": 0.8427,
      "step": 975
    },
    {
      "epoch": 0.2845273668099993,
      "grad_norm": 37.13430703939004,
      "learning_rate": 8.399545388406798e-06,
      "loss": 1.0026,
      "step": 976
    },
    {
      "epoch": 0.284818890751403,
      "grad_norm": 30.126373954047686,
      "learning_rate": 8.39608275845516e-06,
      "loss": 1.232,
      "step": 977
    },
    {
      "epoch": 0.2851104146928066,
      "grad_norm": 13.609425981579049,
      "learning_rate": 8.392617102207945e-06,
      "loss": 0.9949,
      "step": 978
    },
    {
      "epoch": 0.2854019386342103,
      "grad_norm": 15.419574709938415,
      "learning_rate": 8.38914842275345e-06,
      "loss": 0.9843,
      "step": 979
    },
    {
      "epoch": 0.285693462575614,
      "grad_norm": 15.54244321802866,
      "learning_rate": 8.38567672318266e-06,
      "loss": 0.8582,
      "step": 980
    },
    {
      "epoch": 0.2859849865170177,
      "grad_norm": 27.047762326961323,
      "learning_rate": 8.382202006589254e-06,
      "loss": 0.9475,
      "step": 981
    },
    {
      "epoch": 0.2862765104584214,
      "grad_norm": 26.012999488204034,
      "learning_rate": 8.3787242760696e-06,
      "loss": 0.8685,
      "step": 982
    },
    {
      "epoch": 0.2865680343998251,
      "grad_norm": 24.824998619301876,
      "learning_rate": 8.375243534722752e-06,
      "loss": 1.1197,
      "step": 983
    },
    {
      "epoch": 0.2868595583412288,
      "grad_norm": 22.684002786081663,
      "learning_rate": 8.371759785650444e-06,
      "loss": 0.9091,
      "step": 984
    },
    {
      "epoch": 0.2871510822826325,
      "grad_norm": 25.455392679378775,
      "learning_rate": 8.368273031957093e-06,
      "loss": 0.9676,
      "step": 985
    },
    {
      "epoch": 0.28744260622403617,
      "grad_norm": 17.12469035212676,
      "learning_rate": 8.364783276749794e-06,
      "loss": 0.9282,
      "step": 986
    },
    {
      "epoch": 0.28773413016543986,
      "grad_norm": 29.843714493735156,
      "learning_rate": 8.361290523138314e-06,
      "loss": 1.0358,
      "step": 987
    },
    {
      "epoch": 0.2880256541068435,
      "grad_norm": 20.387422262232285,
      "learning_rate": 8.357794774235094e-06,
      "loss": 1.111,
      "step": 988
    },
    {
      "epoch": 0.2883171780482472,
      "grad_norm": 19.82201844715854,
      "learning_rate": 8.354296033155241e-06,
      "loss": 0.8838,
      "step": 989
    },
    {
      "epoch": 0.2886087019896509,
      "grad_norm": 31.57404145408327,
      "learning_rate": 8.350794303016533e-06,
      "loss": 1.0664,
      "step": 990
    },
    {
      "epoch": 0.2889002259310546,
      "grad_norm": 14.168150385647035,
      "learning_rate": 8.34728958693941e-06,
      "loss": 1.0519,
      "step": 991
    },
    {
      "epoch": 0.28919174987245827,
      "grad_norm": 23.60501590321005,
      "learning_rate": 8.343781888046971e-06,
      "loss": 0.8916,
      "step": 992
    },
    {
      "epoch": 0.28948327381386196,
      "grad_norm": 15.53023138142974,
      "learning_rate": 8.340271209464974e-06,
      "loss": 0.879,
      "step": 993
    },
    {
      "epoch": 0.28977479775526566,
      "grad_norm": 23.625259238843892,
      "learning_rate": 8.336757554321832e-06,
      "loss": 0.8186,
      "step": 994
    },
    {
      "epoch": 0.29006632169666935,
      "grad_norm": 111.38372015408333,
      "learning_rate": 8.333240925748615e-06,
      "loss": 0.8583,
      "step": 995
    },
    {
      "epoch": 0.29035784563807304,
      "grad_norm": 22.99840999186663,
      "learning_rate": 8.329721326879032e-06,
      "loss": 1.0938,
      "step": 996
    },
    {
      "epoch": 0.29064936957947674,
      "grad_norm": 20.882043377369644,
      "learning_rate": 8.326198760849452e-06,
      "loss": 1.0026,
      "step": 997
    },
    {
      "epoch": 0.29094089352088043,
      "grad_norm": 20.00454967236962,
      "learning_rate": 8.322673230798877e-06,
      "loss": 0.8989,
      "step": 998
    },
    {
      "epoch": 0.29123241746228407,
      "grad_norm": 21.72479424047989,
      "learning_rate": 8.319144739868958e-06,
      "loss": 0.9398,
      "step": 999
    },
    {
      "epoch": 0.29152394140368776,
      "grad_norm": 22.99957971396452,
      "learning_rate": 8.315613291203977e-06,
      "loss": 0.9174,
      "step": 1000
    },
    {
      "epoch": 0.29152394140368776,
      "eval_loss": 1.1535991430282593,
      "eval_runtime": 2095.4911,
      "eval_samples_per_second": 8.242,
      "eval_steps_per_second": 0.458,
      "step": 1000
    },
    {
      "epoch": 0.29181546534509145,
      "grad_norm": 24.54684112942688,
      "learning_rate": 8.31207888795086e-06,
      "loss": 1.002,
      "step": 1001
    },
    {
      "epoch": 0.29210698928649514,
      "grad_norm": 16.27246071040642,
      "learning_rate": 8.30854153325916e-06,
      "loss": 0.9157,
      "step": 1002
    },
    {
      "epoch": 0.29239851322789884,
      "grad_norm": 25.937637879400132,
      "learning_rate": 8.305001230281062e-06,
      "loss": 1.0131,
      "step": 1003
    },
    {
      "epoch": 0.29269003716930253,
      "grad_norm": 32.98739796357448,
      "learning_rate": 8.30145798217138e-06,
      "loss": 0.9916,
      "step": 1004
    },
    {
      "epoch": 0.2929815611107062,
      "grad_norm": 16.620743363299955,
      "learning_rate": 8.297911792087549e-06,
      "loss": 1.1121,
      "step": 1005
    },
    {
      "epoch": 0.2932730850521099,
      "grad_norm": 16.769584083994886,
      "learning_rate": 8.294362663189626e-06,
      "loss": 0.9456,
      "step": 1006
    },
    {
      "epoch": 0.2935646089935136,
      "grad_norm": 16.765728361918093,
      "learning_rate": 8.290810598640289e-06,
      "loss": 0.9402,
      "step": 1007
    },
    {
      "epoch": 0.2938561329349173,
      "grad_norm": 17.22676208349989,
      "learning_rate": 8.287255601604834e-06,
      "loss": 1.1536,
      "step": 1008
    },
    {
      "epoch": 0.294147656876321,
      "grad_norm": 25.19516498742743,
      "learning_rate": 8.283697675251164e-06,
      "loss": 0.9663,
      "step": 1009
    },
    {
      "epoch": 0.29443918081772463,
      "grad_norm": 25.76647538206754,
      "learning_rate": 8.280136822749796e-06,
      "loss": 0.8588,
      "step": 1010
    },
    {
      "epoch": 0.2947307047591283,
      "grad_norm": 28.734942451804653,
      "learning_rate": 8.276573047273855e-06,
      "loss": 1.1339,
      "step": 1011
    },
    {
      "epoch": 0.295022228700532,
      "grad_norm": 48.73847858421973,
      "learning_rate": 8.27300635199907e-06,
      "loss": 0.8103,
      "step": 1012
    },
    {
      "epoch": 0.2953137526419357,
      "grad_norm": 22.265604090150326,
      "learning_rate": 8.269436740103768e-06,
      "loss": 0.8929,
      "step": 1013
    },
    {
      "epoch": 0.2956052765833394,
      "grad_norm": 29.466522339287433,
      "learning_rate": 8.265864214768883e-06,
      "loss": 0.9923,
      "step": 1014
    },
    {
      "epoch": 0.2958968005247431,
      "grad_norm": 28.757537408844403,
      "learning_rate": 8.262288779177944e-06,
      "loss": 1.0128,
      "step": 1015
    },
    {
      "epoch": 0.2961883244661468,
      "grad_norm": 69.64028289531745,
      "learning_rate": 8.25871043651706e-06,
      "loss": 0.9596,
      "step": 1016
    },
    {
      "epoch": 0.2964798484075505,
      "grad_norm": 16.02837614041517,
      "learning_rate": 8.255129189974953e-06,
      "loss": 1.0009,
      "step": 1017
    },
    {
      "epoch": 0.2967713723489542,
      "grad_norm": 25.825367234105094,
      "learning_rate": 8.25154504274291e-06,
      "loss": 0.9708,
      "step": 1018
    },
    {
      "epoch": 0.29706289629035787,
      "grad_norm": 30.448105635022547,
      "learning_rate": 8.247957998014819e-06,
      "loss": 1.0867,
      "step": 1019
    },
    {
      "epoch": 0.2973544202317615,
      "grad_norm": 24.383882042530075,
      "learning_rate": 8.244368058987145e-06,
      "loss": 1.0668,
      "step": 1020
    },
    {
      "epoch": 0.2976459441731652,
      "grad_norm": 57.82427808954694,
      "learning_rate": 8.240775228858928e-06,
      "loss": 1.2151,
      "step": 1021
    },
    {
      "epoch": 0.2979374681145689,
      "grad_norm": 35.572272513350185,
      "learning_rate": 8.237179510831792e-06,
      "loss": 0.9647,
      "step": 1022
    },
    {
      "epoch": 0.2982289920559726,
      "grad_norm": 15.38547529426313,
      "learning_rate": 8.233580908109925e-06,
      "loss": 0.8973,
      "step": 1023
    },
    {
      "epoch": 0.2985205159973763,
      "grad_norm": 25.36165328345767,
      "learning_rate": 8.229979423900095e-06,
      "loss": 0.94,
      "step": 1024
    },
    {
      "epoch": 0.29881203993877997,
      "grad_norm": 22.30793219863289,
      "learning_rate": 8.226375061411633e-06,
      "loss": 0.9871,
      "step": 1025
    },
    {
      "epoch": 0.29910356388018366,
      "grad_norm": 17.18717002033371,
      "learning_rate": 8.222767823856435e-06,
      "loss": 0.9659,
      "step": 1026
    },
    {
      "epoch": 0.29939508782158736,
      "grad_norm": 20.286402216160976,
      "learning_rate": 8.219157714448957e-06,
      "loss": 1.0882,
      "step": 1027
    },
    {
      "epoch": 0.29968661176299105,
      "grad_norm": 23.413284115790876,
      "learning_rate": 8.215544736406223e-06,
      "loss": 0.8986,
      "step": 1028
    },
    {
      "epoch": 0.29997813570439474,
      "grad_norm": 19.112555948278096,
      "learning_rate": 8.2119288929478e-06,
      "loss": 1.1542,
      "step": 1029
    },
    {
      "epoch": 0.30026965964579844,
      "grad_norm": 74.89627925362497,
      "learning_rate": 8.20831018729582e-06,
      "loss": 0.8434,
      "step": 1030
    },
    {
      "epoch": 0.3005611835872021,
      "grad_norm": 48.82630765031442,
      "learning_rate": 8.204688622674958e-06,
      "loss": 0.9678,
      "step": 1031
    },
    {
      "epoch": 0.30085270752860577,
      "grad_norm": 21.916487023223716,
      "learning_rate": 8.20106420231244e-06,
      "loss": 0.9497,
      "step": 1032
    },
    {
      "epoch": 0.30114423147000946,
      "grad_norm": 45.65463383034681,
      "learning_rate": 8.197436929438039e-06,
      "loss": 1.0502,
      "step": 1033
    },
    {
      "epoch": 0.30143575541141315,
      "grad_norm": 16.396067953937447,
      "learning_rate": 8.193806807284064e-06,
      "loss": 1.1979,
      "step": 1034
    },
    {
      "epoch": 0.30172727935281685,
      "grad_norm": 19.700671689014747,
      "learning_rate": 8.190173839085369e-06,
      "loss": 0.8493,
      "step": 1035
    },
    {
      "epoch": 0.30201880329422054,
      "grad_norm": 19.740079214180597,
      "learning_rate": 8.186538028079338e-06,
      "loss": 0.9675,
      "step": 1036
    },
    {
      "epoch": 0.30231032723562423,
      "grad_norm": 23.919872846243603,
      "learning_rate": 8.182899377505894e-06,
      "loss": 0.9497,
      "step": 1037
    },
    {
      "epoch": 0.3026018511770279,
      "grad_norm": 21.459388269622636,
      "learning_rate": 8.179257890607489e-06,
      "loss": 0.9541,
      "step": 1038
    },
    {
      "epoch": 0.3028933751184316,
      "grad_norm": 26.79177188316781,
      "learning_rate": 8.175613570629099e-06,
      "loss": 1.0296,
      "step": 1039
    },
    {
      "epoch": 0.3031848990598353,
      "grad_norm": 38.55931328824041,
      "learning_rate": 8.171966420818227e-06,
      "loss": 0.9894,
      "step": 1040
    },
    {
      "epoch": 0.303476423001239,
      "grad_norm": 25.290137594353084,
      "learning_rate": 8.168316444424902e-06,
      "loss": 1.0885,
      "step": 1041
    },
    {
      "epoch": 0.30376794694264264,
      "grad_norm": 17.670080916614413,
      "learning_rate": 8.164663644701662e-06,
      "loss": 0.9528,
      "step": 1042
    },
    {
      "epoch": 0.30405947088404633,
      "grad_norm": 20.699913476860416,
      "learning_rate": 8.161008024903572e-06,
      "loss": 1.0371,
      "step": 1043
    },
    {
      "epoch": 0.30435099482545,
      "grad_norm": 45.35163435392413,
      "learning_rate": 8.157349588288202e-06,
      "loss": 0.9162,
      "step": 1044
    },
    {
      "epoch": 0.3046425187668537,
      "grad_norm": 19.798368981150503,
      "learning_rate": 8.153688338115632e-06,
      "loss": 0.9984,
      "step": 1045
    },
    {
      "epoch": 0.3049340427082574,
      "grad_norm": 21.121316399695264,
      "learning_rate": 8.150024277648458e-06,
      "loss": 1.0726,
      "step": 1046
    },
    {
      "epoch": 0.3052255666496611,
      "grad_norm": 38.84949549542965,
      "learning_rate": 8.146357410151769e-06,
      "loss": 1.0108,
      "step": 1047
    },
    {
      "epoch": 0.3055170905910648,
      "grad_norm": 14.320846665893779,
      "learning_rate": 8.142687738893161e-06,
      "loss": 1.0207,
      "step": 1048
    },
    {
      "epoch": 0.3058086145324685,
      "grad_norm": 25.237099657854376,
      "learning_rate": 8.13901526714273e-06,
      "loss": 0.9715,
      "step": 1049
    },
    {
      "epoch": 0.3061001384738722,
      "grad_norm": 28.166042583850498,
      "learning_rate": 8.135339998173064e-06,
      "loss": 0.9745,
      "step": 1050
    },
    {
      "epoch": 0.3061001384738722,
      "eval_loss": 1.1507059335708618,
      "eval_runtime": 2204.8206,
      "eval_samples_per_second": 7.833,
      "eval_steps_per_second": 0.435,
      "step": 1050
    },
    {
      "epoch": 0.3063916624152759,
      "grad_norm": 26.306005977465684,
      "learning_rate": 8.131661935259249e-06,
      "loss": 0.9928,
      "step": 1051
    },
    {
      "epoch": 0.3066831863566795,
      "grad_norm": 12.266581041143809,
      "learning_rate": 8.12798108167885e-06,
      "loss": 1.0373,
      "step": 1052
    },
    {
      "epoch": 0.3069747102980832,
      "grad_norm": 26.130821023973827,
      "learning_rate": 8.124297440711933e-06,
      "loss": 0.939,
      "step": 1053
    },
    {
      "epoch": 0.3072662342394869,
      "grad_norm": 21.61755009774599,
      "learning_rate": 8.120611015641036e-06,
      "loss": 0.8974,
      "step": 1054
    },
    {
      "epoch": 0.3075577581808906,
      "grad_norm": 19.274580322536593,
      "learning_rate": 8.116921809751183e-06,
      "loss": 0.9456,
      "step": 1055
    },
    {
      "epoch": 0.3078492821222943,
      "grad_norm": 35.52247836752489,
      "learning_rate": 8.113229826329876e-06,
      "loss": 0.8656,
      "step": 1056
    },
    {
      "epoch": 0.308140806063698,
      "grad_norm": 29.385088888736796,
      "learning_rate": 8.109535068667094e-06,
      "loss": 1.0186,
      "step": 1057
    },
    {
      "epoch": 0.3084323300051017,
      "grad_norm": 39.855844444997615,
      "learning_rate": 8.105837540055284e-06,
      "loss": 0.9894,
      "step": 1058
    },
    {
      "epoch": 0.30872385394650537,
      "grad_norm": 17.11786091433739,
      "learning_rate": 8.102137243789361e-06,
      "loss": 0.8762,
      "step": 1059
    },
    {
      "epoch": 0.30901537788790906,
      "grad_norm": 24.384297904148593,
      "learning_rate": 8.098434183166716e-06,
      "loss": 0.8906,
      "step": 1060
    },
    {
      "epoch": 0.30930690182931275,
      "grad_norm": 45.287047463852296,
      "learning_rate": 8.094728361487193e-06,
      "loss": 0.9975,
      "step": 1061
    },
    {
      "epoch": 0.30959842577071645,
      "grad_norm": 82.50633175792531,
      "learning_rate": 8.091019782053097e-06,
      "loss": 0.9528,
      "step": 1062
    },
    {
      "epoch": 0.3098899497121201,
      "grad_norm": 23.394376037633382,
      "learning_rate": 8.087308448169198e-06,
      "loss": 1.0369,
      "step": 1063
    },
    {
      "epoch": 0.3101814736535238,
      "grad_norm": 24.576212540433136,
      "learning_rate": 8.083594363142717e-06,
      "loss": 0.9884,
      "step": 1064
    },
    {
      "epoch": 0.31047299759492747,
      "grad_norm": 20.419762429787568,
      "learning_rate": 8.079877530283322e-06,
      "loss": 1.0286,
      "step": 1065
    },
    {
      "epoch": 0.31076452153633116,
      "grad_norm": 47.194325691375674,
      "learning_rate": 8.076157952903134e-06,
      "loss": 0.9178,
      "step": 1066
    },
    {
      "epoch": 0.31105604547773485,
      "grad_norm": 46.68582013566288,
      "learning_rate": 8.072435634316723e-06,
      "loss": 0.8752,
      "step": 1067
    },
    {
      "epoch": 0.31134756941913855,
      "grad_norm": 28.043055237457036,
      "learning_rate": 8.068710577841093e-06,
      "loss": 1.0693,
      "step": 1068
    },
    {
      "epoch": 0.31163909336054224,
      "grad_norm": 42.29754615150238,
      "learning_rate": 8.064982786795695e-06,
      "loss": 1.126,
      "step": 1069
    },
    {
      "epoch": 0.31193061730194593,
      "grad_norm": 29.99966636436443,
      "learning_rate": 8.061252264502415e-06,
      "loss": 1.2502,
      "step": 1070
    },
    {
      "epoch": 0.3122221412433496,
      "grad_norm": 11.875940933398315,
      "learning_rate": 8.057519014285569e-06,
      "loss": 0.9696,
      "step": 1071
    },
    {
      "epoch": 0.3125136651847533,
      "grad_norm": 39.0208041803742,
      "learning_rate": 8.053783039471909e-06,
      "loss": 1.0551,
      "step": 1072
    },
    {
      "epoch": 0.312805189126157,
      "grad_norm": 35.5074287528322,
      "learning_rate": 8.050044343390616e-06,
      "loss": 1.2116,
      "step": 1073
    },
    {
      "epoch": 0.31309671306756065,
      "grad_norm": 22.74384749458522,
      "learning_rate": 8.046302929373286e-06,
      "loss": 0.8819,
      "step": 1074
    },
    {
      "epoch": 0.31338823700896434,
      "grad_norm": 16.738144212022778,
      "learning_rate": 8.04255880075395e-06,
      "loss": 0.9124,
      "step": 1075
    },
    {
      "epoch": 0.31367976095036804,
      "grad_norm": 21.587643862813966,
      "learning_rate": 8.038811960869051e-06,
      "loss": 1.037,
      "step": 1076
    },
    {
      "epoch": 0.31397128489177173,
      "grad_norm": 30.758001743894056,
      "learning_rate": 8.035062413057446e-06,
      "loss": 0.9003,
      "step": 1077
    },
    {
      "epoch": 0.3142628088331754,
      "grad_norm": 17.42869449439322,
      "learning_rate": 8.031310160660411e-06,
      "loss": 1.0235,
      "step": 1078
    },
    {
      "epoch": 0.3145543327745791,
      "grad_norm": 28.134489373905268,
      "learning_rate": 8.02755520702163e-06,
      "loss": 1.1015,
      "step": 1079
    },
    {
      "epoch": 0.3148458567159828,
      "grad_norm": 24.112857949266285,
      "learning_rate": 8.023797555487188e-06,
      "loss": 0.9461,
      "step": 1080
    },
    {
      "epoch": 0.3151373806573865,
      "grad_norm": 78.61931187746103,
      "learning_rate": 8.020037209405585e-06,
      "loss": 0.9185,
      "step": 1081
    },
    {
      "epoch": 0.3154289045987902,
      "grad_norm": 18.36461145618073,
      "learning_rate": 8.016274172127715e-06,
      "loss": 1.005,
      "step": 1082
    },
    {
      "epoch": 0.3157204285401939,
      "grad_norm": 20.314585683601063,
      "learning_rate": 8.012508447006871e-06,
      "loss": 0.8976,
      "step": 1083
    },
    {
      "epoch": 0.3160119524815975,
      "grad_norm": 20.878434276468784,
      "learning_rate": 8.008740037398742e-06,
      "loss": 0.8478,
      "step": 1084
    },
    {
      "epoch": 0.3163034764230012,
      "grad_norm": 18.899273443744,
      "learning_rate": 8.00496894666141e-06,
      "loss": 0.9135,
      "step": 1085
    },
    {
      "epoch": 0.3165950003644049,
      "grad_norm": 14.526725853433605,
      "learning_rate": 8.001195178155344e-06,
      "loss": 0.8168,
      "step": 1086
    },
    {
      "epoch": 0.3168865243058086,
      "grad_norm": 32.87786090170567,
      "learning_rate": 7.9974187352434e-06,
      "loss": 1.029,
      "step": 1087
    },
    {
      "epoch": 0.3171780482472123,
      "grad_norm": 23.989695959245715,
      "learning_rate": 7.99363962129082e-06,
      "loss": 1.0623,
      "step": 1088
    },
    {
      "epoch": 0.317469572188616,
      "grad_norm": 19.207142162615305,
      "learning_rate": 7.989857839665223e-06,
      "loss": 0.9463,
      "step": 1089
    },
    {
      "epoch": 0.3177610961300197,
      "grad_norm": 17.766526713130126,
      "learning_rate": 7.986073393736607e-06,
      "loss": 0.9984,
      "step": 1090
    },
    {
      "epoch": 0.3180526200714234,
      "grad_norm": 30.88534881944116,
      "learning_rate": 7.982286286877342e-06,
      "loss": 1.1392,
      "step": 1091
    },
    {
      "epoch": 0.31834414401282707,
      "grad_norm": 38.91302570156881,
      "learning_rate": 7.978496522462167e-06,
      "loss": 0.9713,
      "step": 1092
    },
    {
      "epoch": 0.31863566795423076,
      "grad_norm": 18.723114102565855,
      "learning_rate": 7.974704103868199e-06,
      "loss": 0.8709,
      "step": 1093
    },
    {
      "epoch": 0.31892719189563445,
      "grad_norm": 27.88655422604864,
      "learning_rate": 7.97090903447491e-06,
      "loss": 0.9727,
      "step": 1094
    },
    {
      "epoch": 0.3192187158370381,
      "grad_norm": 22.30986692896506,
      "learning_rate": 7.96711131766414e-06,
      "loss": 1.0489,
      "step": 1095
    },
    {
      "epoch": 0.3195102397784418,
      "grad_norm": 23.144391883101463,
      "learning_rate": 7.963310956820085e-06,
      "loss": 0.8968,
      "step": 1096
    },
    {
      "epoch": 0.3198017637198455,
      "grad_norm": 22.828131363239354,
      "learning_rate": 7.959507955329298e-06,
      "loss": 0.9319,
      "step": 1097
    },
    {
      "epoch": 0.32009328766124917,
      "grad_norm": 53.739850731906316,
      "learning_rate": 7.955702316580686e-06,
      "loss": 0.8679,
      "step": 1098
    },
    {
      "epoch": 0.32038481160265286,
      "grad_norm": 15.79622996148586,
      "learning_rate": 7.951894043965507e-06,
      "loss": 1.0062,
      "step": 1099
    },
    {
      "epoch": 0.32067633554405656,
      "grad_norm": 30.098736691882998,
      "learning_rate": 7.94808314087736e-06,
      "loss": 1.0652,
      "step": 1100
    },
    {
      "epoch": 0.32067633554405656,
      "eval_loss": 1.143930196762085,
      "eval_runtime": 2032.7947,
      "eval_samples_per_second": 8.496,
      "eval_steps_per_second": 0.472,
      "step": 1100
    },
    {
      "epoch": 0.32096785948546025,
      "grad_norm": 25.03864046760283,
      "learning_rate": 7.944269610712198e-06,
      "loss": 1.0881,
      "step": 1101
    },
    {
      "epoch": 0.32125938342686394,
      "grad_norm": 29.938847822792194,
      "learning_rate": 7.940453456868304e-06,
      "loss": 0.994,
      "step": 1102
    },
    {
      "epoch": 0.32155090736826764,
      "grad_norm": 41.00567582735777,
      "learning_rate": 7.936634682746312e-06,
      "loss": 0.9426,
      "step": 1103
    },
    {
      "epoch": 0.32184243130967133,
      "grad_norm": 21.117643574959448,
      "learning_rate": 7.932813291749177e-06,
      "loss": 0.9284,
      "step": 1104
    },
    {
      "epoch": 0.322133955251075,
      "grad_norm": 18.754699152342788,
      "learning_rate": 7.928989287282195e-06,
      "loss": 0.9701,
      "step": 1105
    },
    {
      "epoch": 0.32242547919247866,
      "grad_norm": 26.297004775624973,
      "learning_rate": 7.925162672752989e-06,
      "loss": 0.9839,
      "step": 1106
    },
    {
      "epoch": 0.32271700313388235,
      "grad_norm": 17.389413180337172,
      "learning_rate": 7.921333451571506e-06,
      "loss": 0.9771,
      "step": 1107
    },
    {
      "epoch": 0.32300852707528604,
      "grad_norm": 25.07523549753182,
      "learning_rate": 7.917501627150019e-06,
      "loss": 0.9645,
      "step": 1108
    },
    {
      "epoch": 0.32330005101668974,
      "grad_norm": 22.07510287140895,
      "learning_rate": 7.913667202903117e-06,
      "loss": 0.8172,
      "step": 1109
    },
    {
      "epoch": 0.32359157495809343,
      "grad_norm": 19.154473616120544,
      "learning_rate": 7.90983018224771e-06,
      "loss": 1.0659,
      "step": 1110
    },
    {
      "epoch": 0.3238830988994971,
      "grad_norm": 15.450424210021058,
      "learning_rate": 7.905990568603014e-06,
      "loss": 0.9098,
      "step": 1111
    },
    {
      "epoch": 0.3241746228409008,
      "grad_norm": 17.162332660288588,
      "learning_rate": 7.902148365390567e-06,
      "loss": 0.9751,
      "step": 1112
    },
    {
      "epoch": 0.3244661467823045,
      "grad_norm": 16.378560009762165,
      "learning_rate": 7.898303576034205e-06,
      "loss": 0.7623,
      "step": 1113
    },
    {
      "epoch": 0.3247576707237082,
      "grad_norm": 21.72536698704497,
      "learning_rate": 7.894456203960075e-06,
      "loss": 1.0945,
      "step": 1114
    },
    {
      "epoch": 0.3250491946651119,
      "grad_norm": 31.529404757810664,
      "learning_rate": 7.89060625259662e-06,
      "loss": 0.8405,
      "step": 1115
    },
    {
      "epoch": 0.32534071860651553,
      "grad_norm": 33.63774106114416,
      "learning_rate": 7.886753725374586e-06,
      "loss": 1.1135,
      "step": 1116
    },
    {
      "epoch": 0.3256322425479192,
      "grad_norm": 25.29961499730671,
      "learning_rate": 7.882898625727011e-06,
      "loss": 1.1304,
      "step": 1117
    },
    {
      "epoch": 0.3259237664893229,
      "grad_norm": 43.94675230935236,
      "learning_rate": 7.879040957089229e-06,
      "loss": 1.1615,
      "step": 1118
    },
    {
      "epoch": 0.3262152904307266,
      "grad_norm": 14.910360370163854,
      "learning_rate": 7.875180722898857e-06,
      "loss": 0.9482,
      "step": 1119
    },
    {
      "epoch": 0.3265068143721303,
      "grad_norm": 51.76194995516656,
      "learning_rate": 7.871317926595804e-06,
      "loss": 0.8717,
      "step": 1120
    },
    {
      "epoch": 0.326798338313534,
      "grad_norm": 20.007899125298945,
      "learning_rate": 7.867452571622262e-06,
      "loss": 0.9651,
      "step": 1121
    },
    {
      "epoch": 0.3270898622549377,
      "grad_norm": 18.936944589577415,
      "learning_rate": 7.8635846614227e-06,
      "loss": 0.9745,
      "step": 1122
    },
    {
      "epoch": 0.3273813861963414,
      "grad_norm": 25.544091511677895,
      "learning_rate": 7.859714199443867e-06,
      "loss": 0.9541,
      "step": 1123
    },
    {
      "epoch": 0.3276729101377451,
      "grad_norm": 42.08837712476309,
      "learning_rate": 7.855841189134784e-06,
      "loss": 0.9121,
      "step": 1124
    },
    {
      "epoch": 0.32796443407914877,
      "grad_norm": 30.37487850022224,
      "learning_rate": 7.85196563394674e-06,
      "loss": 0.9287,
      "step": 1125
    },
    {
      "epoch": 0.32825595802055246,
      "grad_norm": 47.72619251777685,
      "learning_rate": 7.848087537333298e-06,
      "loss": 0.9337,
      "step": 1126
    },
    {
      "epoch": 0.3285474819619561,
      "grad_norm": 45.47187531462484,
      "learning_rate": 7.844206902750283e-06,
      "loss": 0.9438,
      "step": 1127
    },
    {
      "epoch": 0.3288390059033598,
      "grad_norm": 26.016319294268737,
      "learning_rate": 7.84032373365578e-06,
      "loss": 1.0016,
      "step": 1128
    },
    {
      "epoch": 0.3291305298447635,
      "grad_norm": 17.155767902225644,
      "learning_rate": 7.836438033510133e-06,
      "loss": 0.8397,
      "step": 1129
    },
    {
      "epoch": 0.3294220537861672,
      "grad_norm": 20.04562912165317,
      "learning_rate": 7.832549805775945e-06,
      "loss": 0.9535,
      "step": 1130
    },
    {
      "epoch": 0.3297135777275709,
      "grad_norm": 78.02499965083383,
      "learning_rate": 7.828659053918067e-06,
      "loss": 0.8684,
      "step": 1131
    },
    {
      "epoch": 0.33000510166897457,
      "grad_norm": 29.627915948144455,
      "learning_rate": 7.8247657814036e-06,
      "loss": 0.9766,
      "step": 1132
    },
    {
      "epoch": 0.33029662561037826,
      "grad_norm": 37.67145798466422,
      "learning_rate": 7.820869991701894e-06,
      "loss": 1.1127,
      "step": 1133
    },
    {
      "epoch": 0.33058814955178195,
      "grad_norm": 103.27349579611149,
      "learning_rate": 7.81697168828454e-06,
      "loss": 1.0731,
      "step": 1134
    },
    {
      "epoch": 0.33087967349318564,
      "grad_norm": 15.40819651149103,
      "learning_rate": 7.813070874625368e-06,
      "loss": 1.0655,
      "step": 1135
    },
    {
      "epoch": 0.33117119743458934,
      "grad_norm": 19.203524345772767,
      "learning_rate": 7.809167554200446e-06,
      "loss": 0.9017,
      "step": 1136
    },
    {
      "epoch": 0.33146272137599303,
      "grad_norm": 121.83515557308884,
      "learning_rate": 7.805261730488078e-06,
      "loss": 0.7524,
      "step": 1137
    },
    {
      "epoch": 0.33175424531739667,
      "grad_norm": 14.334687222660637,
      "learning_rate": 7.801353406968795e-06,
      "loss": 0.9204,
      "step": 1138
    },
    {
      "epoch": 0.33204576925880036,
      "grad_norm": 17.961871197397056,
      "learning_rate": 7.797442587125359e-06,
      "loss": 1.0152,
      "step": 1139
    },
    {
      "epoch": 0.33233729320020405,
      "grad_norm": 13.749733536526852,
      "learning_rate": 7.793529274442753e-06,
      "loss": 0.9098,
      "step": 1140
    },
    {
      "epoch": 0.33262881714160775,
      "grad_norm": 18.999520035055824,
      "learning_rate": 7.789613472408182e-06,
      "loss": 0.8394,
      "step": 1141
    },
    {
      "epoch": 0.33292034108301144,
      "grad_norm": 18.13593153171338,
      "learning_rate": 7.785695184511074e-06,
      "loss": 0.8367,
      "step": 1142
    },
    {
      "epoch": 0.33321186502441513,
      "grad_norm": 21.573162236799472,
      "learning_rate": 7.781774414243067e-06,
      "loss": 0.8586,
      "step": 1143
    },
    {
      "epoch": 0.3335033889658188,
      "grad_norm": 18.905262818112103,
      "learning_rate": 7.777851165098012e-06,
      "loss": 1.0404,
      "step": 1144
    },
    {
      "epoch": 0.3337949129072225,
      "grad_norm": 16.36969150964594,
      "learning_rate": 7.77392544057197e-06,
      "loss": 0.9776,
      "step": 1145
    },
    {
      "epoch": 0.3340864368486262,
      "grad_norm": 28.665276403269853,
      "learning_rate": 7.769997244163209e-06,
      "loss": 1.1016,
      "step": 1146
    },
    {
      "epoch": 0.3343779607900299,
      "grad_norm": 21.46681330038341,
      "learning_rate": 7.766066579372198e-06,
      "loss": 0.9915,
      "step": 1147
    },
    {
      "epoch": 0.33466948473143354,
      "grad_norm": 13.934324905989902,
      "learning_rate": 7.762133449701603e-06,
      "loss": 1.0023,
      "step": 1148
    },
    {
      "epoch": 0.33496100867283723,
      "grad_norm": 23.342603697041618,
      "learning_rate": 7.758197858656294e-06,
      "loss": 0.9035,
      "step": 1149
    },
    {
      "epoch": 0.33525253261424093,
      "grad_norm": 17.803852356585335,
      "learning_rate": 7.754259809743325e-06,
      "loss": 0.9163,
      "step": 1150
    },
    {
      "epoch": 0.33525253261424093,
      "eval_loss": 1.1437547206878662,
      "eval_runtime": 2074.4286,
      "eval_samples_per_second": 8.325,
      "eval_steps_per_second": 0.463,
      "step": 1150
    },
    {
      "epoch": 0.3355440565556446,
      "grad_norm": 17.27083188553401,
      "learning_rate": 7.75031930647195e-06,
      "loss": 0.9544,
      "step": 1151
    },
    {
      "epoch": 0.3358355804970483,
      "grad_norm": 18.024855250830672,
      "learning_rate": 7.746376352353599e-06,
      "loss": 0.9421,
      "step": 1152
    },
    {
      "epoch": 0.336127104438452,
      "grad_norm": 14.693106823331803,
      "learning_rate": 7.742430950901898e-06,
      "loss": 0.8626,
      "step": 1153
    },
    {
      "epoch": 0.3364186283798557,
      "grad_norm": 22.584720258936365,
      "learning_rate": 7.738483105632644e-06,
      "loss": 0.7932,
      "step": 1154
    },
    {
      "epoch": 0.3367101523212594,
      "grad_norm": 22.78308121390628,
      "learning_rate": 7.734532820063819e-06,
      "loss": 1.0257,
      "step": 1155
    },
    {
      "epoch": 0.3370016762626631,
      "grad_norm": 40.27979863600982,
      "learning_rate": 7.730580097715575e-06,
      "loss": 0.9741,
      "step": 1156
    },
    {
      "epoch": 0.3372932002040668,
      "grad_norm": 94.12593850816839,
      "learning_rate": 7.726624942110233e-06,
      "loss": 0.9352,
      "step": 1157
    },
    {
      "epoch": 0.33758472414547047,
      "grad_norm": 22.948381156213,
      "learning_rate": 7.722667356772291e-06,
      "loss": 0.9624,
      "step": 1158
    },
    {
      "epoch": 0.3378762480868741,
      "grad_norm": 24.895253545836194,
      "learning_rate": 7.718707345228405e-06,
      "loss": 0.8922,
      "step": 1159
    },
    {
      "epoch": 0.3381677720282778,
      "grad_norm": 32.500146125564065,
      "learning_rate": 7.714744911007395e-06,
      "loss": 0.8199,
      "step": 1160
    },
    {
      "epoch": 0.3384592959696815,
      "grad_norm": 65.75303424778393,
      "learning_rate": 7.710780057640239e-06,
      "loss": 1.0138,
      "step": 1161
    },
    {
      "epoch": 0.3387508199110852,
      "grad_norm": 14.859320750562242,
      "learning_rate": 7.706812788660075e-06,
      "loss": 0.8912,
      "step": 1162
    },
    {
      "epoch": 0.3390423438524889,
      "grad_norm": 20.57324594192638,
      "learning_rate": 7.70284310760219e-06,
      "loss": 0.7841,
      "step": 1163
    },
    {
      "epoch": 0.3393338677938926,
      "grad_norm": 22.83844389799892,
      "learning_rate": 7.698871018004016e-06,
      "loss": 0.9468,
      "step": 1164
    },
    {
      "epoch": 0.33962539173529627,
      "grad_norm": 26.385868141761087,
      "learning_rate": 7.694896523405142e-06,
      "loss": 0.8616,
      "step": 1165
    },
    {
      "epoch": 0.33991691567669996,
      "grad_norm": 17.630227813694873,
      "learning_rate": 7.690919627347292e-06,
      "loss": 0.9994,
      "step": 1166
    },
    {
      "epoch": 0.34020843961810365,
      "grad_norm": 26.44015943149855,
      "learning_rate": 7.686940333374333e-06,
      "loss": 0.8719,
      "step": 1167
    },
    {
      "epoch": 0.34049996355950735,
      "grad_norm": 13.141038407550889,
      "learning_rate": 7.682958645032265e-06,
      "loss": 0.8461,
      "step": 1168
    },
    {
      "epoch": 0.34079148750091104,
      "grad_norm": 21.73919248043063,
      "learning_rate": 7.678974565869228e-06,
      "loss": 0.9661,
      "step": 1169
    },
    {
      "epoch": 0.3410830114423147,
      "grad_norm": 15.460931228377854,
      "learning_rate": 7.674988099435487e-06,
      "loss": 0.9091,
      "step": 1170
    },
    {
      "epoch": 0.34137453538371837,
      "grad_norm": 24.265605107508925,
      "learning_rate": 7.670999249283437e-06,
      "loss": 0.9692,
      "step": 1171
    },
    {
      "epoch": 0.34166605932512206,
      "grad_norm": 17.051834287409005,
      "learning_rate": 7.667008018967598e-06,
      "loss": 0.9269,
      "step": 1172
    },
    {
      "epoch": 0.34195758326652576,
      "grad_norm": 24.432216097312327,
      "learning_rate": 7.663014412044611e-06,
      "loss": 0.9437,
      "step": 1173
    },
    {
      "epoch": 0.34224910720792945,
      "grad_norm": 15.870270272202175,
      "learning_rate": 7.65901843207323e-06,
      "loss": 0.8868,
      "step": 1174
    },
    {
      "epoch": 0.34254063114933314,
      "grad_norm": 34.424030221583806,
      "learning_rate": 7.655020082614327e-06,
      "loss": 1.068,
      "step": 1175
    },
    {
      "epoch": 0.34283215509073683,
      "grad_norm": 78.6566629894667,
      "learning_rate": 7.651019367230886e-06,
      "loss": 1.0266,
      "step": 1176
    },
    {
      "epoch": 0.3431236790321405,
      "grad_norm": 24.83045482720967,
      "learning_rate": 7.647016289488003e-06,
      "loss": 1.1044,
      "step": 1177
    },
    {
      "epoch": 0.3434152029735442,
      "grad_norm": 20.21957276753069,
      "learning_rate": 7.643010852952871e-06,
      "loss": 1.0312,
      "step": 1178
    },
    {
      "epoch": 0.3437067269149479,
      "grad_norm": 24.942247096077633,
      "learning_rate": 7.639003061194787e-06,
      "loss": 1.1043,
      "step": 1179
    },
    {
      "epoch": 0.34399825085635155,
      "grad_norm": 23.9945155739898,
      "learning_rate": 7.634992917785156e-06,
      "loss": 0.9915,
      "step": 1180
    },
    {
      "epoch": 0.34428977479775524,
      "grad_norm": 20.020791348729887,
      "learning_rate": 7.630980426297466e-06,
      "loss": 0.7853,
      "step": 1181
    },
    {
      "epoch": 0.34458129873915894,
      "grad_norm": 74.39371201205275,
      "learning_rate": 7.626965590307305e-06,
      "loss": 0.8671,
      "step": 1182
    },
    {
      "epoch": 0.34487282268056263,
      "grad_norm": 18.49858940899121,
      "learning_rate": 7.6229484133923445e-06,
      "loss": 1.0095,
      "step": 1183
    },
    {
      "epoch": 0.3451643466219663,
      "grad_norm": 15.165730865524834,
      "learning_rate": 7.6189288991323505e-06,
      "loss": 1.1837,
      "step": 1184
    },
    {
      "epoch": 0.34545587056337,
      "grad_norm": 19.697460608296822,
      "learning_rate": 7.614907051109164e-06,
      "loss": 1.123,
      "step": 1185
    },
    {
      "epoch": 0.3457473945047737,
      "grad_norm": 21.925546382493952,
      "learning_rate": 7.610882872906709e-06,
      "loss": 0.9547,
      "step": 1186
    },
    {
      "epoch": 0.3460389184461774,
      "grad_norm": 25.906476010860967,
      "learning_rate": 7.606856368110984e-06,
      "loss": 1.052,
      "step": 1187
    },
    {
      "epoch": 0.3463304423875811,
      "grad_norm": 14.680644878255753,
      "learning_rate": 7.602827540310065e-06,
      "loss": 0.9457,
      "step": 1188
    },
    {
      "epoch": 0.3466219663289848,
      "grad_norm": 63.83245000129311,
      "learning_rate": 7.598796393094094e-06,
      "loss": 0.8946,
      "step": 1189
    },
    {
      "epoch": 0.3469134902703885,
      "grad_norm": 31.756607293025745,
      "learning_rate": 7.594762930055281e-06,
      "loss": 0.8793,
      "step": 1190
    },
    {
      "epoch": 0.3472050142117921,
      "grad_norm": 22.581028398359067,
      "learning_rate": 7.5907271547879e-06,
      "loss": 0.9472,
      "step": 1191
    },
    {
      "epoch": 0.3474965381531958,
      "grad_norm": 19.468041873769607,
      "learning_rate": 7.586689070888284e-06,
      "loss": 0.8906,
      "step": 1192
    },
    {
      "epoch": 0.3477880620945995,
      "grad_norm": 25.719861447671065,
      "learning_rate": 7.582648681954827e-06,
      "loss": 0.8367,
      "step": 1193
    },
    {
      "epoch": 0.3480795860360032,
      "grad_norm": 14.890152505403302,
      "learning_rate": 7.578605991587974e-06,
      "loss": 0.9292,
      "step": 1194
    },
    {
      "epoch": 0.3483711099774069,
      "grad_norm": 73.01216298177762,
      "learning_rate": 7.574561003390219e-06,
      "loss": 1.1963,
      "step": 1195
    },
    {
      "epoch": 0.3486626339188106,
      "grad_norm": 30.460737542095618,
      "learning_rate": 7.570513720966108e-06,
      "loss": 1.1909,
      "step": 1196
    },
    {
      "epoch": 0.3489541578602143,
      "grad_norm": 27.699355575747735,
      "learning_rate": 7.566464147922231e-06,
      "loss": 0.8941,
      "step": 1197
    },
    {
      "epoch": 0.34924568180161797,
      "grad_norm": 15.710306830472653,
      "learning_rate": 7.562412287867214e-06,
      "loss": 0.899,
      "step": 1198
    },
    {
      "epoch": 0.34953720574302166,
      "grad_norm": 22.923656486926422,
      "learning_rate": 7.5583581444117284e-06,
      "loss": 0.9291,
      "step": 1199
    },
    {
      "epoch": 0.34982872968442535,
      "grad_norm": 95.41459057085578,
      "learning_rate": 7.5543017211684745e-06,
      "loss": 0.8517,
      "step": 1200
    },
    {
      "epoch": 0.34982872968442535,
      "eval_loss": 1.143210530281067,
      "eval_runtime": 2153.8512,
      "eval_samples_per_second": 8.018,
      "eval_steps_per_second": 0.446,
      "step": 1200
    },
    {
      "epoch": 0.35012025362582905,
      "grad_norm": 15.095846377773919,
      "learning_rate": 7.550243021752186e-06,
      "loss": 0.8283,
      "step": 1201
    },
    {
      "epoch": 0.3504117775672327,
      "grad_norm": 21.542232155004307,
      "learning_rate": 7.5461820497796255e-06,
      "loss": 0.8524,
      "step": 1202
    },
    {
      "epoch": 0.3507033015086364,
      "grad_norm": 34.82512618368011,
      "learning_rate": 7.542118808869582e-06,
      "loss": 0.8327,
      "step": 1203
    },
    {
      "epoch": 0.35099482545004007,
      "grad_norm": 23.14584988648653,
      "learning_rate": 7.5380533026428625e-06,
      "loss": 0.8612,
      "step": 1204
    },
    {
      "epoch": 0.35128634939144376,
      "grad_norm": 19.499195642385807,
      "learning_rate": 7.533985534722294e-06,
      "loss": 0.958,
      "step": 1205
    },
    {
      "epoch": 0.35157787333284746,
      "grad_norm": 22.36567749473638,
      "learning_rate": 7.529915508732725e-06,
      "loss": 0.7953,
      "step": 1206
    },
    {
      "epoch": 0.35186939727425115,
      "grad_norm": 18.182298378725065,
      "learning_rate": 7.525843228301006e-06,
      "loss": 1.0025,
      "step": 1207
    },
    {
      "epoch": 0.35216092121565484,
      "grad_norm": 19.600925907208595,
      "learning_rate": 7.521768697056004e-06,
      "loss": 0.9459,
      "step": 1208
    },
    {
      "epoch": 0.35245244515705854,
      "grad_norm": 27.218842558268253,
      "learning_rate": 7.517691918628589e-06,
      "loss": 0.9498,
      "step": 1209
    },
    {
      "epoch": 0.35274396909846223,
      "grad_norm": 19.86901176750338,
      "learning_rate": 7.513612896651632e-06,
      "loss": 1.1445,
      "step": 1210
    },
    {
      "epoch": 0.3530354930398659,
      "grad_norm": 34.9150806870352,
      "learning_rate": 7.509531634760006e-06,
      "loss": 0.912,
      "step": 1211
    },
    {
      "epoch": 0.35332701698126956,
      "grad_norm": 21.781797924793416,
      "learning_rate": 7.505448136590583e-06,
      "loss": 0.8487,
      "step": 1212
    },
    {
      "epoch": 0.35361854092267325,
      "grad_norm": 87.47257620702028,
      "learning_rate": 7.501362405782217e-06,
      "loss": 1.0024,
      "step": 1213
    },
    {
      "epoch": 0.35391006486407695,
      "grad_norm": 17.99599428606555,
      "learning_rate": 7.497274445975762e-06,
      "loss": 0.844,
      "step": 1214
    },
    {
      "epoch": 0.35420158880548064,
      "grad_norm": 26.70206546414977,
      "learning_rate": 7.493184260814055e-06,
      "loss": 1.0287,
      "step": 1215
    },
    {
      "epoch": 0.35449311274688433,
      "grad_norm": 12.861703652572928,
      "learning_rate": 7.489091853941914e-06,
      "loss": 0.8542,
      "step": 1216
    },
    {
      "epoch": 0.354784636688288,
      "grad_norm": 18.696000369661867,
      "learning_rate": 7.4849972290061405e-06,
      "loss": 1.0267,
      "step": 1217
    },
    {
      "epoch": 0.3550761606296917,
      "grad_norm": 18.668095773249174,
      "learning_rate": 7.480900389655508e-06,
      "loss": 0.9116,
      "step": 1218
    },
    {
      "epoch": 0.3553676845710954,
      "grad_norm": 21.078848879092863,
      "learning_rate": 7.476801339540768e-06,
      "loss": 0.8742,
      "step": 1219
    },
    {
      "epoch": 0.3556592085124991,
      "grad_norm": 28.736327628149123,
      "learning_rate": 7.4727000823146386e-06,
      "loss": 1.0782,
      "step": 1220
    },
    {
      "epoch": 0.3559507324539028,
      "grad_norm": 24.968252131939018,
      "learning_rate": 7.468596621631807e-06,
      "loss": 1.0939,
      "step": 1221
    },
    {
      "epoch": 0.3562422563953065,
      "grad_norm": 102.37059501416542,
      "learning_rate": 7.464490961148921e-06,
      "loss": 0.8056,
      "step": 1222
    },
    {
      "epoch": 0.3565337803367101,
      "grad_norm": 13.690791094153347,
      "learning_rate": 7.460383104524592e-06,
      "loss": 0.9953,
      "step": 1223
    },
    {
      "epoch": 0.3568253042781138,
      "grad_norm": 16.515358125252384,
      "learning_rate": 7.4562730554193875e-06,
      "loss": 0.7846,
      "step": 1224
    },
    {
      "epoch": 0.3571168282195175,
      "grad_norm": 102.26833383875335,
      "learning_rate": 7.4521608174958286e-06,
      "loss": 0.8523,
      "step": 1225
    },
    {
      "epoch": 0.3574083521609212,
      "grad_norm": 22.845670263147507,
      "learning_rate": 7.448046394418383e-06,
      "loss": 0.9777,
      "step": 1226
    },
    {
      "epoch": 0.3576998761023249,
      "grad_norm": 28.951177460767774,
      "learning_rate": 7.443929789853472e-06,
      "loss": 0.9902,
      "step": 1227
    },
    {
      "epoch": 0.3579914000437286,
      "grad_norm": 14.959606845604306,
      "learning_rate": 7.439811007469457e-06,
      "loss": 0.9682,
      "step": 1228
    },
    {
      "epoch": 0.3582829239851323,
      "grad_norm": 18.016646571905664,
      "learning_rate": 7.435690050936642e-06,
      "loss": 0.9289,
      "step": 1229
    },
    {
      "epoch": 0.358574447926536,
      "grad_norm": 20.375609426516448,
      "learning_rate": 7.431566923927267e-06,
      "loss": 0.958,
      "step": 1230
    },
    {
      "epoch": 0.35886597186793967,
      "grad_norm": 19.9423794817202,
      "learning_rate": 7.427441630115507e-06,
      "loss": 0.8208,
      "step": 1231
    },
    {
      "epoch": 0.35915749580934336,
      "grad_norm": 28.13878319208135,
      "learning_rate": 7.423314173177467e-06,
      "loss": 0.764,
      "step": 1232
    },
    {
      "epoch": 0.35944901975074706,
      "grad_norm": 45.042177006118884,
      "learning_rate": 7.419184556791179e-06,
      "loss": 0.9755,
      "step": 1233
    },
    {
      "epoch": 0.3597405436921507,
      "grad_norm": 74.6261980176809,
      "learning_rate": 7.415052784636603e-06,
      "loss": 1.1054,
      "step": 1234
    },
    {
      "epoch": 0.3600320676335544,
      "grad_norm": 21.610451344267062,
      "learning_rate": 7.410918860395615e-06,
      "loss": 0.8545,
      "step": 1235
    },
    {
      "epoch": 0.3603235915749581,
      "grad_norm": 20.69725511407702,
      "learning_rate": 7.406782787752011e-06,
      "loss": 0.8929,
      "step": 1236
    },
    {
      "epoch": 0.3606151155163618,
      "grad_norm": 29.734371610528523,
      "learning_rate": 7.402644570391504e-06,
      "loss": 0.9954,
      "step": 1237
    },
    {
      "epoch": 0.36090663945776547,
      "grad_norm": 31.46489624909901,
      "learning_rate": 7.398504212001714e-06,
      "loss": 1.0188,
      "step": 1238
    },
    {
      "epoch": 0.36119816339916916,
      "grad_norm": 44.29938110056778,
      "learning_rate": 7.3943617162721715e-06,
      "loss": 0.7566,
      "step": 1239
    },
    {
      "epoch": 0.36148968734057285,
      "grad_norm": 26.262264354049165,
      "learning_rate": 7.390217086894309e-06,
      "loss": 0.7738,
      "step": 1240
    },
    {
      "epoch": 0.36178121128197654,
      "grad_norm": 23.624521749450096,
      "learning_rate": 7.386070327561465e-06,
      "loss": 0.8743,
      "step": 1241
    },
    {
      "epoch": 0.36207273522338024,
      "grad_norm": 27.904184910928162,
      "learning_rate": 7.3819214419688725e-06,
      "loss": 0.6666,
      "step": 1242
    },
    {
      "epoch": 0.36236425916478393,
      "grad_norm": 53.16891417832891,
      "learning_rate": 7.377770433813658e-06,
      "loss": 1.0113,
      "step": 1243
    },
    {
      "epoch": 0.36265578310618757,
      "grad_norm": 28.152380142781084,
      "learning_rate": 7.373617306794844e-06,
      "loss": 0.9706,
      "step": 1244
    },
    {
      "epoch": 0.36294730704759126,
      "grad_norm": 33.48044953909447,
      "learning_rate": 7.369462064613336e-06,
      "loss": 0.8907,
      "step": 1245
    },
    {
      "epoch": 0.36323883098899495,
      "grad_norm": 20.519897981227817,
      "learning_rate": 7.365304710971928e-06,
      "loss": 0.9997,
      "step": 1246
    },
    {
      "epoch": 0.36353035493039865,
      "grad_norm": 27.324041457754163,
      "learning_rate": 7.3611452495752945e-06,
      "loss": 0.8418,
      "step": 1247
    },
    {
      "epoch": 0.36382187887180234,
      "grad_norm": 15.308775942367252,
      "learning_rate": 7.3569836841299905e-06,
      "loss": 0.8313,
      "step": 1248
    },
    {
      "epoch": 0.36411340281320603,
      "grad_norm": 22.82641252398897,
      "learning_rate": 7.352820018344437e-06,
      "loss": 0.926,
      "step": 1249
    },
    {
      "epoch": 0.3644049267546097,
      "grad_norm": 29.2206347818303,
      "learning_rate": 7.348654255928941e-06,
      "loss": 0.8613,
      "step": 1250
    },
    {
      "epoch": 0.3644049267546097,
      "eval_loss": 1.145037293434143,
      "eval_runtime": 2197.9226,
      "eval_samples_per_second": 7.857,
      "eval_steps_per_second": 0.437,
      "step": 1250
    },
    {
      "epoch": 0.3646964506960134,
      "grad_norm": 28.521695967952414,
      "learning_rate": 7.344486400595665e-06,
      "loss": 0.9439,
      "step": 1251
    },
    {
      "epoch": 0.3649879746374171,
      "grad_norm": 23.383025246845396,
      "learning_rate": 7.340316456058644e-06,
      "loss": 0.9618,
      "step": 1252
    },
    {
      "epoch": 0.3652794985788208,
      "grad_norm": 21.6568578011528,
      "learning_rate": 7.336144426033771e-06,
      "loss": 1.007,
      "step": 1253
    },
    {
      "epoch": 0.3655710225202245,
      "grad_norm": 47.4993893720625,
      "learning_rate": 7.331970314238799e-06,
      "loss": 0.9744,
      "step": 1254
    },
    {
      "epoch": 0.36586254646162814,
      "grad_norm": 92.23320241394548,
      "learning_rate": 7.327794124393337e-06,
      "loss": 0.8433,
      "step": 1255
    },
    {
      "epoch": 0.36615407040303183,
      "grad_norm": 91.00008645484297,
      "learning_rate": 7.323615860218844e-06,
      "loss": 0.9195,
      "step": 1256
    },
    {
      "epoch": 0.3664455943444355,
      "grad_norm": 35.76061736354705,
      "learning_rate": 7.319435525438627e-06,
      "loss": 1.0132,
      "step": 1257
    },
    {
      "epoch": 0.3667371182858392,
      "grad_norm": 28.508867317964906,
      "learning_rate": 7.31525312377784e-06,
      "loss": 0.9409,
      "step": 1258
    },
    {
      "epoch": 0.3670286422272429,
      "grad_norm": 20.9506424479957,
      "learning_rate": 7.3110686589634794e-06,
      "loss": 0.8192,
      "step": 1259
    },
    {
      "epoch": 0.3673201661686466,
      "grad_norm": 36.51681358733153,
      "learning_rate": 7.306882134724376e-06,
      "loss": 0.886,
      "step": 1260
    },
    {
      "epoch": 0.3676116901100503,
      "grad_norm": 17.669596435041566,
      "learning_rate": 7.3026935547912004e-06,
      "loss": 0.986,
      "step": 1261
    },
    {
      "epoch": 0.367903214051454,
      "grad_norm": 20.236332982580603,
      "learning_rate": 7.298502922896453e-06,
      "loss": 0.8989,
      "step": 1262
    },
    {
      "epoch": 0.3681947379928577,
      "grad_norm": 24.44223482475648,
      "learning_rate": 7.294310242774464e-06,
      "loss": 0.8026,
      "step": 1263
    },
    {
      "epoch": 0.36848626193426137,
      "grad_norm": 20.2168997979298,
      "learning_rate": 7.290115518161385e-06,
      "loss": 0.9398,
      "step": 1264
    },
    {
      "epoch": 0.36877778587566507,
      "grad_norm": 14.794050428718519,
      "learning_rate": 7.285918752795195e-06,
      "loss": 0.9042,
      "step": 1265
    },
    {
      "epoch": 0.3690693098170687,
      "grad_norm": 14.586922631371761,
      "learning_rate": 7.281719950415686e-06,
      "loss": 1.0149,
      "step": 1266
    },
    {
      "epoch": 0.3693608337584724,
      "grad_norm": 16.820401121098804,
      "learning_rate": 7.27751911476447e-06,
      "loss": 1.1968,
      "step": 1267
    },
    {
      "epoch": 0.3696523576998761,
      "grad_norm": 26.797523271830283,
      "learning_rate": 7.273316249584969e-06,
      "loss": 0.9093,
      "step": 1268
    },
    {
      "epoch": 0.3699438816412798,
      "grad_norm": 22.59171481498091,
      "learning_rate": 7.269111358622412e-06,
      "loss": 0.9665,
      "step": 1269
    },
    {
      "epoch": 0.3702354055826835,
      "grad_norm": 21.637743827600197,
      "learning_rate": 7.2649044456238334e-06,
      "loss": 0.8453,
      "step": 1270
    },
    {
      "epoch": 0.37052692952408717,
      "grad_norm": 22.98295074974201,
      "learning_rate": 7.260695514338074e-06,
      "loss": 1.1045,
      "step": 1271
    },
    {
      "epoch": 0.37081845346549086,
      "grad_norm": 20.159317101760603,
      "learning_rate": 7.256484568515769e-06,
      "loss": 0.8578,
      "step": 1272
    },
    {
      "epoch": 0.37110997740689455,
      "grad_norm": 20.87964556328915,
      "learning_rate": 7.252271611909347e-06,
      "loss": 0.7399,
      "step": 1273
    },
    {
      "epoch": 0.37140150134829825,
      "grad_norm": 14.595396769361114,
      "learning_rate": 7.248056648273034e-06,
      "loss": 0.7703,
      "step": 1274
    },
    {
      "epoch": 0.37169302528970194,
      "grad_norm": 17.520637929542204,
      "learning_rate": 7.243839681362839e-06,
      "loss": 0.8556,
      "step": 1275
    },
    {
      "epoch": 0.3719845492311056,
      "grad_norm": 26.66958728527831,
      "learning_rate": 7.239620714936561e-06,
      "loss": 0.997,
      "step": 1276
    },
    {
      "epoch": 0.37227607317250927,
      "grad_norm": 14.784204609125215,
      "learning_rate": 7.2353997527537765e-06,
      "loss": 0.8898,
      "step": 1277
    },
    {
      "epoch": 0.37256759711391296,
      "grad_norm": 22.45308357743358,
      "learning_rate": 7.231176798575843e-06,
      "loss": 0.8677,
      "step": 1278
    },
    {
      "epoch": 0.37285912105531666,
      "grad_norm": 17.833585727543394,
      "learning_rate": 7.226951856165893e-06,
      "loss": 0.8881,
      "step": 1279
    },
    {
      "epoch": 0.37315064499672035,
      "grad_norm": 114.47662509822202,
      "learning_rate": 7.22272492928883e-06,
      "loss": 1.1126,
      "step": 1280
    },
    {
      "epoch": 0.37344216893812404,
      "grad_norm": 31.94104661315291,
      "learning_rate": 7.218496021711325e-06,
      "loss": 0.8879,
      "step": 1281
    },
    {
      "epoch": 0.37373369287952773,
      "grad_norm": 23.09202788731205,
      "learning_rate": 7.214265137201817e-06,
      "loss": 1.0328,
      "step": 1282
    },
    {
      "epoch": 0.3740252168209314,
      "grad_norm": 21.43296161725837,
      "learning_rate": 7.210032279530503e-06,
      "loss": 0.8789,
      "step": 1283
    },
    {
      "epoch": 0.3743167407623351,
      "grad_norm": 20.79424136292502,
      "learning_rate": 7.205797452469341e-06,
      "loss": 0.9865,
      "step": 1284
    },
    {
      "epoch": 0.3746082647037388,
      "grad_norm": 16.300589145451198,
      "learning_rate": 7.201560659792041e-06,
      "loss": 0.859,
      "step": 1285
    },
    {
      "epoch": 0.3748997886451425,
      "grad_norm": 22.100133609212484,
      "learning_rate": 7.197321905274071e-06,
      "loss": 0.9189,
      "step": 1286
    },
    {
      "epoch": 0.37519131258654614,
      "grad_norm": 22.522847636548825,
      "learning_rate": 7.193081192692639e-06,
      "loss": 0.8908,
      "step": 1287
    },
    {
      "epoch": 0.37548283652794984,
      "grad_norm": 21.237736844325624,
      "learning_rate": 7.188838525826702e-06,
      "loss": 0.848,
      "step": 1288
    },
    {
      "epoch": 0.37577436046935353,
      "grad_norm": 47.78057326588442,
      "learning_rate": 7.184593908456961e-06,
      "loss": 0.7588,
      "step": 1289
    },
    {
      "epoch": 0.3760658844107572,
      "grad_norm": 26.80478905523514,
      "learning_rate": 7.18034734436585e-06,
      "loss": 0.7592,
      "step": 1290
    },
    {
      "epoch": 0.3763574083521609,
      "grad_norm": 19.45310658518201,
      "learning_rate": 7.176098837337541e-06,
      "loss": 0.8395,
      "step": 1291
    },
    {
      "epoch": 0.3766489322935646,
      "grad_norm": 30.909185812979203,
      "learning_rate": 7.171848391157935e-06,
      "loss": 0.8673,
      "step": 1292
    },
    {
      "epoch": 0.3769404562349683,
      "grad_norm": 30.429073628896543,
      "learning_rate": 7.167596009614664e-06,
      "loss": 0.9375,
      "step": 1293
    },
    {
      "epoch": 0.377231980176372,
      "grad_norm": 17.529589048352975,
      "learning_rate": 7.163341696497084e-06,
      "loss": 0.8262,
      "step": 1294
    },
    {
      "epoch": 0.3775235041177757,
      "grad_norm": 25.720710821522612,
      "learning_rate": 7.159085455596267e-06,
      "loss": 0.6809,
      "step": 1295
    },
    {
      "epoch": 0.3778150280591794,
      "grad_norm": 38.1582487028113,
      "learning_rate": 7.154827290705012e-06,
      "loss": 1.0238,
      "step": 1296
    },
    {
      "epoch": 0.3781065520005831,
      "grad_norm": 21.639912796274942,
      "learning_rate": 7.150567205617825e-06,
      "loss": 0.892,
      "step": 1297
    },
    {
      "epoch": 0.3783980759419867,
      "grad_norm": 16.460236991649207,
      "learning_rate": 7.146305204130928e-06,
      "loss": 0.7695,
      "step": 1298
    },
    {
      "epoch": 0.3786895998833904,
      "grad_norm": 17.911352506564146,
      "learning_rate": 7.142041290042245e-06,
      "loss": 1.0314,
      "step": 1299
    },
    {
      "epoch": 0.3789811238247941,
      "grad_norm": 15.757503977620766,
      "learning_rate": 7.137775467151411e-06,
      "loss": 1.0331,
      "step": 1300
    },
    {
      "epoch": 0.3789811238247941,
      "eval_loss": 1.1423577070236206,
      "eval_runtime": 2215.4648,
      "eval_samples_per_second": 7.795,
      "eval_steps_per_second": 0.433,
      "step": 1300
    },
    {
      "epoch": 0.3792726477661978,
      "grad_norm": 41.01396953762889,
      "learning_rate": 7.1335077392597585e-06,
      "loss": 1.0209,
      "step": 1301
    },
    {
      "epoch": 0.3795641717076015,
      "grad_norm": 29.031063905814424,
      "learning_rate": 7.129238110170315e-06,
      "loss": 0.7933,
      "step": 1302
    },
    {
      "epoch": 0.3798556956490052,
      "grad_norm": 20.436459614072643,
      "learning_rate": 7.124966583687811e-06,
      "loss": 0.9579,
      "step": 1303
    },
    {
      "epoch": 0.38014721959040887,
      "grad_norm": 24.944787643730006,
      "learning_rate": 7.120693163618656e-06,
      "loss": 0.9565,
      "step": 1304
    },
    {
      "epoch": 0.38043874353181256,
      "grad_norm": 27.132988428113357,
      "learning_rate": 7.116417853770955e-06,
      "loss": 0.8889,
      "step": 1305
    },
    {
      "epoch": 0.38073026747321626,
      "grad_norm": 14.525388389452132,
      "learning_rate": 7.112140657954495e-06,
      "loss": 0.7424,
      "step": 1306
    },
    {
      "epoch": 0.38102179141461995,
      "grad_norm": 15.292290467276318,
      "learning_rate": 7.1078615799807445e-06,
      "loss": 0.8017,
      "step": 1307
    },
    {
      "epoch": 0.38131331535602364,
      "grad_norm": 14.97337769533739,
      "learning_rate": 7.103580623662845e-06,
      "loss": 1.1133,
      "step": 1308
    },
    {
      "epoch": 0.3816048392974273,
      "grad_norm": 19.104612329075557,
      "learning_rate": 7.099297792815618e-06,
      "loss": 0.9323,
      "step": 1309
    },
    {
      "epoch": 0.38189636323883097,
      "grad_norm": 33.01204854403518,
      "learning_rate": 7.0950130912555515e-06,
      "loss": 1.0204,
      "step": 1310
    },
    {
      "epoch": 0.38218788718023466,
      "grad_norm": 16.07175706324995,
      "learning_rate": 7.090726522800802e-06,
      "loss": 0.9371,
      "step": 1311
    },
    {
      "epoch": 0.38247941112163836,
      "grad_norm": 46.72226831125824,
      "learning_rate": 7.086438091271186e-06,
      "loss": 0.9155,
      "step": 1312
    },
    {
      "epoch": 0.38277093506304205,
      "grad_norm": 19.250471552367603,
      "learning_rate": 7.0821478004881875e-06,
      "loss": 0.7491,
      "step": 1313
    },
    {
      "epoch": 0.38306245900444574,
      "grad_norm": 30.43743109420279,
      "learning_rate": 7.077855654274939e-06,
      "loss": 0.8168,
      "step": 1314
    },
    {
      "epoch": 0.38335398294584944,
      "grad_norm": 24.50441499104726,
      "learning_rate": 7.0735616564562324e-06,
      "loss": 0.8319,
      "step": 1315
    },
    {
      "epoch": 0.38364550688725313,
      "grad_norm": 14.128871340130127,
      "learning_rate": 7.069265810858509e-06,
      "loss": 0.7669,
      "step": 1316
    },
    {
      "epoch": 0.3839370308286568,
      "grad_norm": 43.10515575542401,
      "learning_rate": 7.064968121309852e-06,
      "loss": 0.7763,
      "step": 1317
    },
    {
      "epoch": 0.3842285547700605,
      "grad_norm": 20.46682941835637,
      "learning_rate": 7.0606685916399945e-06,
      "loss": 0.9678,
      "step": 1318
    },
    {
      "epoch": 0.38452007871146415,
      "grad_norm": 18.799113872404064,
      "learning_rate": 7.056367225680303e-06,
      "loss": 0.8912,
      "step": 1319
    },
    {
      "epoch": 0.38481160265286785,
      "grad_norm": 14.200214946214864,
      "learning_rate": 7.052064027263785e-06,
      "loss": 0.761,
      "step": 1320
    },
    {
      "epoch": 0.38510312659427154,
      "grad_norm": 15.78186223841283,
      "learning_rate": 7.047759000225079e-06,
      "loss": 0.8726,
      "step": 1321
    },
    {
      "epoch": 0.38539465053567523,
      "grad_norm": 15.81847979173775,
      "learning_rate": 7.043452148400452e-06,
      "loss": 0.9332,
      "step": 1322
    },
    {
      "epoch": 0.3856861744770789,
      "grad_norm": 20.935970812150295,
      "learning_rate": 7.039143475627799e-06,
      "loss": 0.8159,
      "step": 1323
    },
    {
      "epoch": 0.3859776984184826,
      "grad_norm": 35.052403852807316,
      "learning_rate": 7.034832985746638e-06,
      "loss": 0.7628,
      "step": 1324
    },
    {
      "epoch": 0.3862692223598863,
      "grad_norm": 24.85190490197069,
      "learning_rate": 7.0305206825981045e-06,
      "loss": 0.8202,
      "step": 1325
    },
    {
      "epoch": 0.38656074630129,
      "grad_norm": 20.64258359738684,
      "learning_rate": 7.026206570024949e-06,
      "loss": 0.736,
      "step": 1326
    },
    {
      "epoch": 0.3868522702426937,
      "grad_norm": 15.345371271617331,
      "learning_rate": 7.021890651871539e-06,
      "loss": 0.9226,
      "step": 1327
    },
    {
      "epoch": 0.3871437941840974,
      "grad_norm": 22.145892461443495,
      "learning_rate": 7.017572931983846e-06,
      "loss": 0.8347,
      "step": 1328
    },
    {
      "epoch": 0.3874353181255011,
      "grad_norm": 80.90517866075693,
      "learning_rate": 7.013253414209448e-06,
      "loss": 0.8359,
      "step": 1329
    },
    {
      "epoch": 0.3877268420669047,
      "grad_norm": 22.053190842635832,
      "learning_rate": 7.00893210239753e-06,
      "loss": 0.9466,
      "step": 1330
    },
    {
      "epoch": 0.3880183660083084,
      "grad_norm": 52.74959140971859,
      "learning_rate": 7.004609000398866e-06,
      "loss": 0.935,
      "step": 1331
    },
    {
      "epoch": 0.3883098899497121,
      "grad_norm": 36.46821011366523,
      "learning_rate": 7.000284112065836e-06,
      "loss": 0.9661,
      "step": 1332
    },
    {
      "epoch": 0.3886014138911158,
      "grad_norm": 22.030750631580922,
      "learning_rate": 6.9959574412524045e-06,
      "loss": 1.0512,
      "step": 1333
    },
    {
      "epoch": 0.3888929378325195,
      "grad_norm": 25.742154908148827,
      "learning_rate": 6.9916289918141265e-06,
      "loss": 0.7963,
      "step": 1334
    },
    {
      "epoch": 0.3891844617739232,
      "grad_norm": 17.40052914298978,
      "learning_rate": 6.987298767608144e-06,
      "loss": 1.0088,
      "step": 1335
    },
    {
      "epoch": 0.3894759857153269,
      "grad_norm": 19.13915946356808,
      "learning_rate": 6.982966772493176e-06,
      "loss": 1.0122,
      "step": 1336
    },
    {
      "epoch": 0.38976750965673057,
      "grad_norm": 28.12941703279665,
      "learning_rate": 6.978633010329526e-06,
      "loss": 0.9269,
      "step": 1337
    },
    {
      "epoch": 0.39005903359813426,
      "grad_norm": 14.061026025624946,
      "learning_rate": 6.974297484979066e-06,
      "loss": 0.9393,
      "step": 1338
    },
    {
      "epoch": 0.39035055753953796,
      "grad_norm": 14.683052834289345,
      "learning_rate": 6.969960200305242e-06,
      "loss": 0.9096,
      "step": 1339
    },
    {
      "epoch": 0.39064208148094165,
      "grad_norm": 14.509269139068822,
      "learning_rate": 6.965621160173066e-06,
      "loss": 0.8961,
      "step": 1340
    },
    {
      "epoch": 0.3909336054223453,
      "grad_norm": 16.1725094063197,
      "learning_rate": 6.961280368449118e-06,
      "loss": 1.0687,
      "step": 1341
    },
    {
      "epoch": 0.391225129363749,
      "grad_norm": 25.7921251887952,
      "learning_rate": 6.9569378290015375e-06,
      "loss": 0.8183,
      "step": 1342
    },
    {
      "epoch": 0.3915166533051527,
      "grad_norm": 30.438243455070978,
      "learning_rate": 6.952593545700017e-06,
      "loss": 0.9778,
      "step": 1343
    },
    {
      "epoch": 0.39180817724655637,
      "grad_norm": 17.34567119386563,
      "learning_rate": 6.948247522415811e-06,
      "loss": 0.9986,
      "step": 1344
    },
    {
      "epoch": 0.39209970118796006,
      "grad_norm": 17.475896555900476,
      "learning_rate": 6.943899763021715e-06,
      "loss": 1.0363,
      "step": 1345
    },
    {
      "epoch": 0.39239122512936375,
      "grad_norm": 18.35565263919456,
      "learning_rate": 6.939550271392079e-06,
      "loss": 0.9738,
      "step": 1346
    },
    {
      "epoch": 0.39268274907076745,
      "grad_norm": 17.242144189039163,
      "learning_rate": 6.935199051402795e-06,
      "loss": 1.0057,
      "step": 1347
    },
    {
      "epoch": 0.39297427301217114,
      "grad_norm": 20.69473508288607,
      "learning_rate": 6.930846106931292e-06,
      "loss": 0.8753,
      "step": 1348
    },
    {
      "epoch": 0.39326579695357483,
      "grad_norm": 18.623055577180292,
      "learning_rate": 6.926491441856541e-06,
      "loss": 0.6414,
      "step": 1349
    },
    {
      "epoch": 0.3935573208949785,
      "grad_norm": 42.801648255223796,
      "learning_rate": 6.922135060059043e-06,
      "loss": 0.9479,
      "step": 1350
    },
    {
      "epoch": 0.3935573208949785,
      "eval_loss": 1.138645052909851,
      "eval_runtime": 2194.468,
      "eval_samples_per_second": 7.87,
      "eval_steps_per_second": 0.437,
      "step": 1350
    },
    {
      "epoch": 0.39384884483638216,
      "grad_norm": 13.70463731613978,
      "learning_rate": 6.917776965420826e-06,
      "loss": 0.8361,
      "step": 1351
    },
    {
      "epoch": 0.39414036877778585,
      "grad_norm": 143.4019983230656,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.9765,
      "step": 1352
    },
    {
      "epoch": 0.39443189271918955,
      "grad_norm": 22.79775075167724,
      "learning_rate": 6.909055653157994e-06,
      "loss": 0.9187,
      "step": 1353
    },
    {
      "epoch": 0.39472341666059324,
      "grad_norm": 36.00125736442532,
      "learning_rate": 6.904692443305059e-06,
      "loss": 0.7503,
      "step": 1354
    },
    {
      "epoch": 0.39501494060199693,
      "grad_norm": 25.730893928718707,
      "learning_rate": 6.900327536154762e-06,
      "loss": 1.0005,
      "step": 1355
    },
    {
      "epoch": 0.3953064645434006,
      "grad_norm": 22.24512540370614,
      "learning_rate": 6.895960935596728e-06,
      "loss": 0.9494,
      "step": 1356
    },
    {
      "epoch": 0.3955979884848043,
      "grad_norm": 20.747091564024178,
      "learning_rate": 6.891592645522095e-06,
      "loss": 0.8491,
      "step": 1357
    },
    {
      "epoch": 0.395889512426208,
      "grad_norm": 23.007517332187444,
      "learning_rate": 6.8872226698235065e-06,
      "loss": 0.9241,
      "step": 1358
    },
    {
      "epoch": 0.3961810363676117,
      "grad_norm": 46.04380581270415,
      "learning_rate": 6.882851012395109e-06,
      "loss": 0.8391,
      "step": 1359
    },
    {
      "epoch": 0.3964725603090154,
      "grad_norm": 21.883146983106194,
      "learning_rate": 6.8784776771325426e-06,
      "loss": 1.0239,
      "step": 1360
    },
    {
      "epoch": 0.3967640842504191,
      "grad_norm": 22.9952146521739,
      "learning_rate": 6.87410266793295e-06,
      "loss": 0.8632,
      "step": 1361
    },
    {
      "epoch": 0.39705560819182273,
      "grad_norm": 44.45504090593887,
      "learning_rate": 6.869725988694955e-06,
      "loss": 1.0073,
      "step": 1362
    },
    {
      "epoch": 0.3973471321332264,
      "grad_norm": 18.07147308601503,
      "learning_rate": 6.865347643318683e-06,
      "loss": 0.8069,
      "step": 1363
    },
    {
      "epoch": 0.3976386560746301,
      "grad_norm": 14.980403057520236,
      "learning_rate": 6.860967635705732e-06,
      "loss": 0.7551,
      "step": 1364
    },
    {
      "epoch": 0.3979301800160338,
      "grad_norm": 24.972720734006046,
      "learning_rate": 6.856585969759189e-06,
      "loss": 0.7353,
      "step": 1365
    },
    {
      "epoch": 0.3982217039574375,
      "grad_norm": 298.4738875077809,
      "learning_rate": 6.8522026493836144e-06,
      "loss": 0.7789,
      "step": 1366
    },
    {
      "epoch": 0.3985132278988412,
      "grad_norm": 15.664699781328853,
      "learning_rate": 6.847817678485048e-06,
      "loss": 1.0262,
      "step": 1367
    },
    {
      "epoch": 0.3988047518402449,
      "grad_norm": 90.83351081580057,
      "learning_rate": 6.843431060970995e-06,
      "loss": 0.8522,
      "step": 1368
    },
    {
      "epoch": 0.3990962757816486,
      "grad_norm": 22.800185813342384,
      "learning_rate": 6.839042800750429e-06,
      "loss": 0.9082,
      "step": 1369
    },
    {
      "epoch": 0.3993877997230523,
      "grad_norm": 26.454992468098354,
      "learning_rate": 6.834652901733789e-06,
      "loss": 0.8926,
      "step": 1370
    },
    {
      "epoch": 0.39967932366445597,
      "grad_norm": 53.82974006683346,
      "learning_rate": 6.830261367832975e-06,
      "loss": 0.8363,
      "step": 1371
    },
    {
      "epoch": 0.39997084760585966,
      "grad_norm": 21.873286135650723,
      "learning_rate": 6.825868202961343e-06,
      "loss": 0.7723,
      "step": 1372
    },
    {
      "epoch": 0.4002623715472633,
      "grad_norm": 18.581829490329067,
      "learning_rate": 6.821473411033699e-06,
      "loss": 0.7909,
      "step": 1373
    },
    {
      "epoch": 0.400553895488667,
      "grad_norm": 33.13318324226744,
      "learning_rate": 6.8170769959663045e-06,
      "loss": 0.8412,
      "step": 1374
    },
    {
      "epoch": 0.4008454194300707,
      "grad_norm": 28.225579688693557,
      "learning_rate": 6.812678961676864e-06,
      "loss": 0.761,
      "step": 1375
    },
    {
      "epoch": 0.4011369433714744,
      "grad_norm": 33.05817986008733,
      "learning_rate": 6.808279312084525e-06,
      "loss": 0.7958,
      "step": 1376
    },
    {
      "epoch": 0.40142846731287807,
      "grad_norm": 37.31871694581517,
      "learning_rate": 6.803878051109878e-06,
      "loss": 0.7753,
      "step": 1377
    },
    {
      "epoch": 0.40171999125428176,
      "grad_norm": 13.865042845881971,
      "learning_rate": 6.799475182674942e-06,
      "loss": 1.0286,
      "step": 1378
    },
    {
      "epoch": 0.40201151519568545,
      "grad_norm": 59.258382060926564,
      "learning_rate": 6.795070710703176e-06,
      "loss": 0.916,
      "step": 1379
    },
    {
      "epoch": 0.40230303913708915,
      "grad_norm": 28.388912889483297,
      "learning_rate": 6.790664639119464e-06,
      "loss": 0.8239,
      "step": 1380
    },
    {
      "epoch": 0.40259456307849284,
      "grad_norm": 17.95014614583471,
      "learning_rate": 6.786256971850118e-06,
      "loss": 0.9094,
      "step": 1381
    },
    {
      "epoch": 0.40288608701989653,
      "grad_norm": 25.67852188424191,
      "learning_rate": 6.781847712822869e-06,
      "loss": 0.755,
      "step": 1382
    },
    {
      "epoch": 0.40317761096130017,
      "grad_norm": 47.788175843836974,
      "learning_rate": 6.777436865966867e-06,
      "loss": 0.803,
      "step": 1383
    },
    {
      "epoch": 0.40346913490270386,
      "grad_norm": 23.836966929647577,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.8688,
      "step": 1384
    },
    {
      "epoch": 0.40376065884410756,
      "grad_norm": 36.45475191885949,
      "learning_rate": 6.7686104244922816e-06,
      "loss": 0.9819,
      "step": 1385
    },
    {
      "epoch": 0.40405218278551125,
      "grad_norm": 27.077391894384824,
      "learning_rate": 6.76419483773906e-06,
      "loss": 0.9778,
      "step": 1386
    },
    {
      "epoch": 0.40434370672691494,
      "grad_norm": 24.839435722278278,
      "learning_rate": 6.759777678887802e-06,
      "loss": 0.8801,
      "step": 1387
    },
    {
      "epoch": 0.40463523066831864,
      "grad_norm": 24.543568037918323,
      "learning_rate": 6.755358951874701e-06,
      "loss": 0.9459,
      "step": 1388
    },
    {
      "epoch": 0.40492675460972233,
      "grad_norm": 32.10750022544483,
      "learning_rate": 6.750938660637342e-06,
      "loss": 0.8465,
      "step": 1389
    },
    {
      "epoch": 0.405218278551126,
      "grad_norm": 10.951720423965563,
      "learning_rate": 6.7465168091147094e-06,
      "loss": 0.7455,
      "step": 1390
    },
    {
      "epoch": 0.4055098024925297,
      "grad_norm": 25.72086532600459,
      "learning_rate": 6.742093401247173e-06,
      "loss": 0.912,
      "step": 1391
    },
    {
      "epoch": 0.4058013264339334,
      "grad_norm": 22.661680838326227,
      "learning_rate": 6.737668440976494e-06,
      "loss": 0.8722,
      "step": 1392
    },
    {
      "epoch": 0.4060928503753371,
      "grad_norm": 19.473863282124714,
      "learning_rate": 6.733241932245811e-06,
      "loss": 0.8115,
      "step": 1393
    },
    {
      "epoch": 0.40638437431674074,
      "grad_norm": 38.7215996922976,
      "learning_rate": 6.728813878999652e-06,
      "loss": 0.9406,
      "step": 1394
    },
    {
      "epoch": 0.40667589825814443,
      "grad_norm": 42.75959191116194,
      "learning_rate": 6.724384285183908e-06,
      "loss": 0.8661,
      "step": 1395
    },
    {
      "epoch": 0.4069674221995481,
      "grad_norm": 18.973002553624145,
      "learning_rate": 6.719953154745857e-06,
      "loss": 0.815,
      "step": 1396
    },
    {
      "epoch": 0.4072589461409518,
      "grad_norm": 16.842087008666372,
      "learning_rate": 6.715520491634136e-06,
      "loss": 0.797,
      "step": 1397
    },
    {
      "epoch": 0.4075504700823555,
      "grad_norm": 20.45004789285521,
      "learning_rate": 6.7110862997987525e-06,
      "loss": 0.848,
      "step": 1398
    },
    {
      "epoch": 0.4078419940237592,
      "grad_norm": 28.910445289580107,
      "learning_rate": 6.706650583191074e-06,
      "loss": 0.9266,
      "step": 1399
    },
    {
      "epoch": 0.4081335179651629,
      "grad_norm": 15.284850219407875,
      "learning_rate": 6.70221334576383e-06,
      "loss": 0.9956,
      "step": 1400
    },
    {
      "epoch": 0.4081335179651629,
      "eval_loss": 1.1402138471603394,
      "eval_runtime": 2288.5606,
      "eval_samples_per_second": 7.546,
      "eval_steps_per_second": 0.419,
      "step": 1400
    },
    {
      "epoch": 0.4084250419065666,
      "grad_norm": 55.67701566211607,
      "learning_rate": 6.697774591471103e-06,
      "loss": 0.9887,
      "step": 1401
    },
    {
      "epoch": 0.4087165658479703,
      "grad_norm": 16.542288814924724,
      "learning_rate": 6.693334324268328e-06,
      "loss": 0.8815,
      "step": 1402
    },
    {
      "epoch": 0.409008089789374,
      "grad_norm": 24.039873638990155,
      "learning_rate": 6.6888925481122845e-06,
      "loss": 0.9589,
      "step": 1403
    },
    {
      "epoch": 0.40929961373077767,
      "grad_norm": 99.60229480734104,
      "learning_rate": 6.684449266961101e-06,
      "loss": 0.8643,
      "step": 1404
    },
    {
      "epoch": 0.4095911376721813,
      "grad_norm": 16.765553300242196,
      "learning_rate": 6.680004484774246e-06,
      "loss": 0.9178,
      "step": 1405
    },
    {
      "epoch": 0.409882661613585,
      "grad_norm": 20.51652334450236,
      "learning_rate": 6.675558205512527e-06,
      "loss": 0.9255,
      "step": 1406
    },
    {
      "epoch": 0.4101741855549887,
      "grad_norm": 42.121385627171044,
      "learning_rate": 6.671110433138081e-06,
      "loss": 0.8724,
      "step": 1407
    },
    {
      "epoch": 0.4104657094963924,
      "grad_norm": 20.65107449740164,
      "learning_rate": 6.666661171614382e-06,
      "loss": 0.8635,
      "step": 1408
    },
    {
      "epoch": 0.4107572334377961,
      "grad_norm": 21.16398950759977,
      "learning_rate": 6.662210424906226e-06,
      "loss": 1.0426,
      "step": 1409
    },
    {
      "epoch": 0.41104875737919977,
      "grad_norm": 20.21177217111699,
      "learning_rate": 6.657758196979732e-06,
      "loss": 0.9994,
      "step": 1410
    },
    {
      "epoch": 0.41134028132060346,
      "grad_norm": 17.368706875157418,
      "learning_rate": 6.653304491802344e-06,
      "loss": 1.0106,
      "step": 1411
    },
    {
      "epoch": 0.41163180526200716,
      "grad_norm": 29.02460072255405,
      "learning_rate": 6.648849313342816e-06,
      "loss": 1.0758,
      "step": 1412
    },
    {
      "epoch": 0.41192332920341085,
      "grad_norm": 47.203303094569215,
      "learning_rate": 6.644392665571221e-06,
      "loss": 0.8704,
      "step": 1413
    },
    {
      "epoch": 0.41221485314481454,
      "grad_norm": 28.34374581552618,
      "learning_rate": 6.6399345524589366e-06,
      "loss": 0.9248,
      "step": 1414
    },
    {
      "epoch": 0.4125063770862182,
      "grad_norm": 21.23229929709708,
      "learning_rate": 6.635474977978651e-06,
      "loss": 0.8655,
      "step": 1415
    },
    {
      "epoch": 0.41279790102762187,
      "grad_norm": 20.747952834981817,
      "learning_rate": 6.631013946104348e-06,
      "loss": 0.8559,
      "step": 1416
    },
    {
      "epoch": 0.41308942496902556,
      "grad_norm": 31.57108542948411,
      "learning_rate": 6.626551460811316e-06,
      "loss": 0.8237,
      "step": 1417
    },
    {
      "epoch": 0.41338094891042926,
      "grad_norm": 29.34015063895226,
      "learning_rate": 6.622087526076135e-06,
      "loss": 0.8886,
      "step": 1418
    },
    {
      "epoch": 0.41367247285183295,
      "grad_norm": 23.086645250050847,
      "learning_rate": 6.61762214587668e-06,
      "loss": 0.8352,
      "step": 1419
    },
    {
      "epoch": 0.41396399679323664,
      "grad_norm": 47.59571965901115,
      "learning_rate": 6.613155324192111e-06,
      "loss": 0.8471,
      "step": 1420
    },
    {
      "epoch": 0.41425552073464034,
      "grad_norm": 19.84045623818977,
      "learning_rate": 6.608687065002873e-06,
      "loss": 0.995,
      "step": 1421
    },
    {
      "epoch": 0.41454704467604403,
      "grad_norm": 14.355474114596454,
      "learning_rate": 6.604217372290693e-06,
      "loss": 0.766,
      "step": 1422
    },
    {
      "epoch": 0.4148385686174477,
      "grad_norm": 27.116267117695887,
      "learning_rate": 6.599746250038577e-06,
      "loss": 0.9227,
      "step": 1423
    },
    {
      "epoch": 0.4151300925588514,
      "grad_norm": 21.35780542533381,
      "learning_rate": 6.5952737022308e-06,
      "loss": 0.982,
      "step": 1424
    },
    {
      "epoch": 0.4154216165002551,
      "grad_norm": 31.40494871985433,
      "learning_rate": 6.590799732852913e-06,
      "loss": 0.7815,
      "step": 1425
    },
    {
      "epoch": 0.41571314044165875,
      "grad_norm": 17.54901729634207,
      "learning_rate": 6.586324345891727e-06,
      "loss": 0.8432,
      "step": 1426
    },
    {
      "epoch": 0.41600466438306244,
      "grad_norm": 21.133279601604674,
      "learning_rate": 6.581847545335323e-06,
      "loss": 0.9114,
      "step": 1427
    },
    {
      "epoch": 0.41629618832446613,
      "grad_norm": 18.72852793040178,
      "learning_rate": 6.57736933517304e-06,
      "loss": 1.0456,
      "step": 1428
    },
    {
      "epoch": 0.4165877122658698,
      "grad_norm": 27.596460545221813,
      "learning_rate": 6.5728897193954685e-06,
      "loss": 0.981,
      "step": 1429
    },
    {
      "epoch": 0.4168792362072735,
      "grad_norm": 27.236604960082907,
      "learning_rate": 6.568408701994459e-06,
      "loss": 0.8272,
      "step": 1430
    },
    {
      "epoch": 0.4171707601486772,
      "grad_norm": 34.044866898313046,
      "learning_rate": 6.563926286963103e-06,
      "loss": 0.9648,
      "step": 1431
    },
    {
      "epoch": 0.4174622840900809,
      "grad_norm": 22.218790842763376,
      "learning_rate": 6.559442478295745e-06,
      "loss": 0.881,
      "step": 1432
    },
    {
      "epoch": 0.4177538080314846,
      "grad_norm": 19.750089995767908,
      "learning_rate": 6.554957279987968e-06,
      "loss": 0.8818,
      "step": 1433
    },
    {
      "epoch": 0.4180453319728883,
      "grad_norm": 16.240241878717512,
      "learning_rate": 6.550470696036591e-06,
      "loss": 0.9235,
      "step": 1434
    },
    {
      "epoch": 0.418336855914292,
      "grad_norm": 20.41604523094361,
      "learning_rate": 6.5459827304396706e-06,
      "loss": 0.8516,
      "step": 1435
    },
    {
      "epoch": 0.4186283798556957,
      "grad_norm": 27.283173397550808,
      "learning_rate": 6.541493387196496e-06,
      "loss": 0.8043,
      "step": 1436
    },
    {
      "epoch": 0.4189199037970993,
      "grad_norm": 23.835305946593984,
      "learning_rate": 6.53700267030758e-06,
      "loss": 0.859,
      "step": 1437
    },
    {
      "epoch": 0.419211427738503,
      "grad_norm": 166.0944017873692,
      "learning_rate": 6.5325105837746604e-06,
      "loss": 0.7475,
      "step": 1438
    },
    {
      "epoch": 0.4195029516799067,
      "grad_norm": 19.551037038380226,
      "learning_rate": 6.5280171316007e-06,
      "loss": 0.8653,
      "step": 1439
    },
    {
      "epoch": 0.4197944756213104,
      "grad_norm": 41.78275892650452,
      "learning_rate": 6.523522317789874e-06,
      "loss": 0.9418,
      "step": 1440
    },
    {
      "epoch": 0.4200859995627141,
      "grad_norm": 23.011233281611744,
      "learning_rate": 6.519026146347571e-06,
      "loss": 0.8354,
      "step": 1441
    },
    {
      "epoch": 0.4203775235041178,
      "grad_norm": 20.4162407514665,
      "learning_rate": 6.514528621280391e-06,
      "loss": 1.0863,
      "step": 1442
    },
    {
      "epoch": 0.42066904744552147,
      "grad_norm": 21.86474605261369,
      "learning_rate": 6.510029746596141e-06,
      "loss": 0.8565,
      "step": 1443
    },
    {
      "epoch": 0.42096057138692516,
      "grad_norm": 23.762293304504784,
      "learning_rate": 6.5055295263038286e-06,
      "loss": 0.9183,
      "step": 1444
    },
    {
      "epoch": 0.42125209532832886,
      "grad_norm": 13.977050204392174,
      "learning_rate": 6.50102796441366e-06,
      "loss": 0.7892,
      "step": 1445
    },
    {
      "epoch": 0.42154361926973255,
      "grad_norm": 47.483084336811,
      "learning_rate": 6.496525064937042e-06,
      "loss": 1.0347,
      "step": 1446
    },
    {
      "epoch": 0.4218351432111362,
      "grad_norm": 18.228942294029896,
      "learning_rate": 6.492020831886565e-06,
      "loss": 0.9364,
      "step": 1447
    },
    {
      "epoch": 0.4221266671525399,
      "grad_norm": 33.72243332295137,
      "learning_rate": 6.487515269276015e-06,
      "loss": 0.8816,
      "step": 1448
    },
    {
      "epoch": 0.4224181910939436,
      "grad_norm": 40.3601771178695,
      "learning_rate": 6.48300838112036e-06,
      "loss": 0.8413,
      "step": 1449
    },
    {
      "epoch": 0.42270971503534727,
      "grad_norm": 22.292515187257607,
      "learning_rate": 6.478500171435751e-06,
      "loss": 1.0062,
      "step": 1450
    },
    {
      "epoch": 0.42270971503534727,
      "eval_loss": 1.1348731517791748,
      "eval_runtime": 2408.8169,
      "eval_samples_per_second": 7.169,
      "eval_steps_per_second": 0.399,
      "step": 1450
    },
    {
      "epoch": 0.42300123897675096,
      "grad_norm": 17.04381056279633,
      "learning_rate": 6.473990644239512e-06,
      "loss": 0.7982,
      "step": 1451
    },
    {
      "epoch": 0.42329276291815465,
      "grad_norm": 21.803721993658478,
      "learning_rate": 6.469479803550144e-06,
      "loss": 1.0485,
      "step": 1452
    },
    {
      "epoch": 0.42358428685955835,
      "grad_norm": 29.004455224290204,
      "learning_rate": 6.4649676533873204e-06,
      "loss": 0.8859,
      "step": 1453
    },
    {
      "epoch": 0.42387581080096204,
      "grad_norm": 15.2011560893895,
      "learning_rate": 6.460454197771881e-06,
      "loss": 0.7256,
      "step": 1454
    },
    {
      "epoch": 0.42416733474236573,
      "grad_norm": 26.22417160941632,
      "learning_rate": 6.455939440725824e-06,
      "loss": 0.8263,
      "step": 1455
    },
    {
      "epoch": 0.4244588586837694,
      "grad_norm": 13.478132092077876,
      "learning_rate": 6.451423386272312e-06,
      "loss": 0.8227,
      "step": 1456
    },
    {
      "epoch": 0.4247503826251731,
      "grad_norm": 17.549504878715187,
      "learning_rate": 6.446906038435666e-06,
      "loss": 0.8518,
      "step": 1457
    },
    {
      "epoch": 0.42504190656657675,
      "grad_norm": 34.23850160066107,
      "learning_rate": 6.442387401241349e-06,
      "loss": 0.9866,
      "step": 1458
    },
    {
      "epoch": 0.42533343050798045,
      "grad_norm": 23.616991083568273,
      "learning_rate": 6.4378674787159866e-06,
      "loss": 1.0419,
      "step": 1459
    },
    {
      "epoch": 0.42562495444938414,
      "grad_norm": 169.11711527479113,
      "learning_rate": 6.433346274887341e-06,
      "loss": 0.7657,
      "step": 1460
    },
    {
      "epoch": 0.42591647839078783,
      "grad_norm": 30.762539752056153,
      "learning_rate": 6.428823793784318e-06,
      "loss": 1.0136,
      "step": 1461
    },
    {
      "epoch": 0.4262080023321915,
      "grad_norm": 16.36898586954753,
      "learning_rate": 6.4243000394369626e-06,
      "loss": 0.9516,
      "step": 1462
    },
    {
      "epoch": 0.4264995262735952,
      "grad_norm": 38.159224175375954,
      "learning_rate": 6.419775015876455e-06,
      "loss": 0.8905,
      "step": 1463
    },
    {
      "epoch": 0.4267910502149989,
      "grad_norm": 29.997346297696012,
      "learning_rate": 6.415248727135103e-06,
      "loss": 0.7972,
      "step": 1464
    },
    {
      "epoch": 0.4270825741564026,
      "grad_norm": 17.3849917706392,
      "learning_rate": 6.4107211772463455e-06,
      "loss": 0.8705,
      "step": 1465
    },
    {
      "epoch": 0.4273740980978063,
      "grad_norm": 26.086224296826305,
      "learning_rate": 6.406192370244742e-06,
      "loss": 0.8154,
      "step": 1466
    },
    {
      "epoch": 0.42766562203921,
      "grad_norm": 16.99736967183641,
      "learning_rate": 6.401662310165979e-06,
      "loss": 0.7996,
      "step": 1467
    },
    {
      "epoch": 0.4279571459806137,
      "grad_norm": 28.039311540142364,
      "learning_rate": 6.397131001046849e-06,
      "loss": 0.8092,
      "step": 1468
    },
    {
      "epoch": 0.4282486699220173,
      "grad_norm": 20.71596810165782,
      "learning_rate": 6.392598446925266e-06,
      "loss": 0.8257,
      "step": 1469
    },
    {
      "epoch": 0.428540193863421,
      "grad_norm": 24.590016668723035,
      "learning_rate": 6.38806465184025e-06,
      "loss": 0.8793,
      "step": 1470
    },
    {
      "epoch": 0.4288317178048247,
      "grad_norm": 17.022892322390966,
      "learning_rate": 6.3835296198319294e-06,
      "loss": 1.1063,
      "step": 1471
    },
    {
      "epoch": 0.4291232417462284,
      "grad_norm": 29.44386536465664,
      "learning_rate": 6.378993354941529e-06,
      "loss": 0.7949,
      "step": 1472
    },
    {
      "epoch": 0.4294147656876321,
      "grad_norm": 17.303139543867413,
      "learning_rate": 6.374455861211382e-06,
      "loss": 0.8778,
      "step": 1473
    },
    {
      "epoch": 0.4297062896290358,
      "grad_norm": 30.85247743951813,
      "learning_rate": 6.3699171426849036e-06,
      "loss": 0.9129,
      "step": 1474
    },
    {
      "epoch": 0.4299978135704395,
      "grad_norm": 32.95301038835499,
      "learning_rate": 6.365377203406613e-06,
      "loss": 0.9041,
      "step": 1475
    },
    {
      "epoch": 0.4302893375118432,
      "grad_norm": 27.35387217379624,
      "learning_rate": 6.3608360474221106e-06,
      "loss": 0.8387,
      "step": 1476
    },
    {
      "epoch": 0.43058086145324687,
      "grad_norm": 23.720319857245393,
      "learning_rate": 6.356293678778082e-06,
      "loss": 1.0598,
      "step": 1477
    },
    {
      "epoch": 0.43087238539465056,
      "grad_norm": 31.98228722506458,
      "learning_rate": 6.3517501015222924e-06,
      "loss": 0.7278,
      "step": 1478
    },
    {
      "epoch": 0.4311639093360542,
      "grad_norm": 21.971009801653146,
      "learning_rate": 6.347205319703587e-06,
      "loss": 0.8071,
      "step": 1479
    },
    {
      "epoch": 0.4314554332774579,
      "grad_norm": 24.045480460055742,
      "learning_rate": 6.342659337371884e-06,
      "loss": 0.9126,
      "step": 1480
    },
    {
      "epoch": 0.4317469572188616,
      "grad_norm": 26.574022161020974,
      "learning_rate": 6.338112158578169e-06,
      "loss": 0.8974,
      "step": 1481
    },
    {
      "epoch": 0.4320384811602653,
      "grad_norm": 26.62400796802811,
      "learning_rate": 6.333563787374493e-06,
      "loss": 0.8623,
      "step": 1482
    },
    {
      "epoch": 0.43233000510166897,
      "grad_norm": 19.262279243185585,
      "learning_rate": 6.329014227813972e-06,
      "loss": 0.8312,
      "step": 1483
    },
    {
      "epoch": 0.43262152904307266,
      "grad_norm": 41.2181485569462,
      "learning_rate": 6.3244634839507834e-06,
      "loss": 0.9098,
      "step": 1484
    },
    {
      "epoch": 0.43291305298447635,
      "grad_norm": 47.111199801511034,
      "learning_rate": 6.319911559840155e-06,
      "loss": 0.8864,
      "step": 1485
    },
    {
      "epoch": 0.43320457692588005,
      "grad_norm": 22.86748314480545,
      "learning_rate": 6.315358459538367e-06,
      "loss": 0.8614,
      "step": 1486
    },
    {
      "epoch": 0.43349610086728374,
      "grad_norm": 32.52341923599634,
      "learning_rate": 6.310804187102752e-06,
      "loss": 0.8895,
      "step": 1487
    },
    {
      "epoch": 0.43378762480868743,
      "grad_norm": 34.05674983815606,
      "learning_rate": 6.3062487465916825e-06,
      "loss": 1.0107,
      "step": 1488
    },
    {
      "epoch": 0.4340791487500911,
      "grad_norm": 26.53920286055712,
      "learning_rate": 6.3016921420645725e-06,
      "loss": 0.9776,
      "step": 1489
    },
    {
      "epoch": 0.43437067269149476,
      "grad_norm": 19.524498137702693,
      "learning_rate": 6.297134377581877e-06,
      "loss": 1.0575,
      "step": 1490
    },
    {
      "epoch": 0.43466219663289846,
      "grad_norm": 71.17033824037095,
      "learning_rate": 6.292575457205079e-06,
      "loss": 0.9744,
      "step": 1491
    },
    {
      "epoch": 0.43495372057430215,
      "grad_norm": 18.59903701885909,
      "learning_rate": 6.2880153849966966e-06,
      "loss": 0.799,
      "step": 1492
    },
    {
      "epoch": 0.43524524451570584,
      "grad_norm": 60.24269938999878,
      "learning_rate": 6.283454165020272e-06,
      "loss": 0.7866,
      "step": 1493
    },
    {
      "epoch": 0.43553676845710954,
      "grad_norm": 24.06997000500635,
      "learning_rate": 6.2788918013403695e-06,
      "loss": 0.8329,
      "step": 1494
    },
    {
      "epoch": 0.43582829239851323,
      "grad_norm": 23.95352109473528,
      "learning_rate": 6.274328298022574e-06,
      "loss": 0.9244,
      "step": 1495
    },
    {
      "epoch": 0.4361198163399169,
      "grad_norm": 27.55950254418215,
      "learning_rate": 6.269763659133486e-06,
      "loss": 0.9097,
      "step": 1496
    },
    {
      "epoch": 0.4364113402813206,
      "grad_norm": 17.690542329508233,
      "learning_rate": 6.265197888740715e-06,
      "loss": 0.7655,
      "step": 1497
    },
    {
      "epoch": 0.4367028642227243,
      "grad_norm": 15.016484552394477,
      "learning_rate": 6.2606309909128845e-06,
      "loss": 0.8712,
      "step": 1498
    },
    {
      "epoch": 0.436994388164128,
      "grad_norm": 22.069284019528858,
      "learning_rate": 6.256062969719617e-06,
      "loss": 0.8239,
      "step": 1499
    },
    {
      "epoch": 0.4372859121055317,
      "grad_norm": 24.302750270945666,
      "learning_rate": 6.251493829231539e-06,
      "loss": 0.8164,
      "step": 1500
    },
    {
      "epoch": 0.4372859121055317,
      "eval_loss": 1.1378263235092163,
      "eval_runtime": 2386.5927,
      "eval_samples_per_second": 7.236,
      "eval_steps_per_second": 0.402,
      "step": 1500
    },
    {
      "epoch": 0.43757743604693533,
      "grad_norm": 55.10602569732308,
      "learning_rate": 6.246923573520274e-06,
      "loss": 0.7267,
      "step": 1501
    },
    {
      "epoch": 0.437868959988339,
      "grad_norm": 28.33022175774175,
      "learning_rate": 6.24235220665844e-06,
      "loss": 0.8609,
      "step": 1502
    },
    {
      "epoch": 0.4381604839297427,
      "grad_norm": 40.54270665762467,
      "learning_rate": 6.237779732719642e-06,
      "loss": 0.9236,
      "step": 1503
    },
    {
      "epoch": 0.4384520078711464,
      "grad_norm": 23.851271028291237,
      "learning_rate": 6.233206155778476e-06,
      "loss": 0.97,
      "step": 1504
    },
    {
      "epoch": 0.4387435318125501,
      "grad_norm": 24.81524097635084,
      "learning_rate": 6.2286314799105176e-06,
      "loss": 1.0154,
      "step": 1505
    },
    {
      "epoch": 0.4390350557539538,
      "grad_norm": 25.089928271684343,
      "learning_rate": 6.224055709192323e-06,
      "loss": 0.8103,
      "step": 1506
    },
    {
      "epoch": 0.4393265796953575,
      "grad_norm": 18.911247427882387,
      "learning_rate": 6.219478847701422e-06,
      "loss": 0.7921,
      "step": 1507
    },
    {
      "epoch": 0.4396181036367612,
      "grad_norm": 20.00016100289918,
      "learning_rate": 6.21490089951632e-06,
      "loss": 0.8367,
      "step": 1508
    },
    {
      "epoch": 0.4399096275781649,
      "grad_norm": 26.713092067923025,
      "learning_rate": 6.210321868716489e-06,
      "loss": 0.9867,
      "step": 1509
    },
    {
      "epoch": 0.44020115151956857,
      "grad_norm": 23.64411014177821,
      "learning_rate": 6.205741759382365e-06,
      "loss": 0.9499,
      "step": 1510
    },
    {
      "epoch": 0.4404926754609722,
      "grad_norm": 38.43971822461267,
      "learning_rate": 6.201160575595349e-06,
      "loss": 1.1863,
      "step": 1511
    },
    {
      "epoch": 0.4407841994023759,
      "grad_norm": 20.02113298583475,
      "learning_rate": 6.1965783214377895e-06,
      "loss": 0.82,
      "step": 1512
    },
    {
      "epoch": 0.4410757233437796,
      "grad_norm": 23.60951133493246,
      "learning_rate": 6.191995000993001e-06,
      "loss": 0.8686,
      "step": 1513
    },
    {
      "epoch": 0.4413672472851833,
      "grad_norm": 34.50531441312375,
      "learning_rate": 6.187410618345241e-06,
      "loss": 0.8878,
      "step": 1514
    },
    {
      "epoch": 0.441658771226587,
      "grad_norm": 14.145535727259603,
      "learning_rate": 6.182825177579716e-06,
      "loss": 0.8894,
      "step": 1515
    },
    {
      "epoch": 0.44195029516799067,
      "grad_norm": 21.191667865841918,
      "learning_rate": 6.178238682782574e-06,
      "loss": 0.9286,
      "step": 1516
    },
    {
      "epoch": 0.44224181910939436,
      "grad_norm": 26.496913137064826,
      "learning_rate": 6.173651138040902e-06,
      "loss": 0.9093,
      "step": 1517
    },
    {
      "epoch": 0.44253334305079806,
      "grad_norm": 32.34121114802119,
      "learning_rate": 6.169062547442724e-06,
      "loss": 0.8592,
      "step": 1518
    },
    {
      "epoch": 0.44282486699220175,
      "grad_norm": 32.38572652454959,
      "learning_rate": 6.164472915076999e-06,
      "loss": 0.8749,
      "step": 1519
    },
    {
      "epoch": 0.44311639093360544,
      "grad_norm": 16.849696107696328,
      "learning_rate": 6.159882245033606e-06,
      "loss": 0.7914,
      "step": 1520
    },
    {
      "epoch": 0.44340791487500913,
      "grad_norm": 20.018799530005893,
      "learning_rate": 6.155290541403357e-06,
      "loss": 0.7525,
      "step": 1521
    },
    {
      "epoch": 0.4436994388164128,
      "grad_norm": 18.7282675447236,
      "learning_rate": 6.150697808277979e-06,
      "loss": 0.8597,
      "step": 1522
    },
    {
      "epoch": 0.44399096275781647,
      "grad_norm": 40.56251778187383,
      "learning_rate": 6.14610404975012e-06,
      "loss": 0.8739,
      "step": 1523
    },
    {
      "epoch": 0.44428248669922016,
      "grad_norm": 32.6011883352564,
      "learning_rate": 6.141509269913343e-06,
      "loss": 0.9473,
      "step": 1524
    },
    {
      "epoch": 0.44457401064062385,
      "grad_norm": 21.534293711675666,
      "learning_rate": 6.136913472862115e-06,
      "loss": 0.8431,
      "step": 1525
    },
    {
      "epoch": 0.44486553458202754,
      "grad_norm": 20.883261047800058,
      "learning_rate": 6.132316662691815e-06,
      "loss": 0.8471,
      "step": 1526
    },
    {
      "epoch": 0.44515705852343124,
      "grad_norm": 46.36610412376142,
      "learning_rate": 6.127718843498723e-06,
      "loss": 0.9259,
      "step": 1527
    },
    {
      "epoch": 0.44544858246483493,
      "grad_norm": 25.83471346282061,
      "learning_rate": 6.123120019380021e-06,
      "loss": 0.9212,
      "step": 1528
    },
    {
      "epoch": 0.4457401064062386,
      "grad_norm": 25.472235621074788,
      "learning_rate": 6.118520194433779e-06,
      "loss": 0.9281,
      "step": 1529
    },
    {
      "epoch": 0.4460316303476423,
      "grad_norm": 16.44781660559995,
      "learning_rate": 6.1139193727589665e-06,
      "loss": 0.9196,
      "step": 1530
    },
    {
      "epoch": 0.446323154289046,
      "grad_norm": 21.85047646200784,
      "learning_rate": 6.109317558455437e-06,
      "loss": 0.9289,
      "step": 1531
    },
    {
      "epoch": 0.4466146782304497,
      "grad_norm": 51.78139868614843,
      "learning_rate": 6.1047147556239325e-06,
      "loss": 0.8362,
      "step": 1532
    },
    {
      "epoch": 0.44690620217185334,
      "grad_norm": 19.836862799403676,
      "learning_rate": 6.100110968366071e-06,
      "loss": 0.7526,
      "step": 1533
    },
    {
      "epoch": 0.44719772611325703,
      "grad_norm": 24.02993251945147,
      "learning_rate": 6.095506200784349e-06,
      "loss": 0.7988,
      "step": 1534
    },
    {
      "epoch": 0.4474892500546607,
      "grad_norm": 26.388556574526348,
      "learning_rate": 6.090900456982139e-06,
      "loss": 0.982,
      "step": 1535
    },
    {
      "epoch": 0.4477807739960644,
      "grad_norm": 19.342541468562434,
      "learning_rate": 6.086293741063685e-06,
      "loss": 0.8947,
      "step": 1536
    },
    {
      "epoch": 0.4480722979374681,
      "grad_norm": 62.377359954724064,
      "learning_rate": 6.081686057134088e-06,
      "loss": 0.9425,
      "step": 1537
    },
    {
      "epoch": 0.4483638218788718,
      "grad_norm": 26.773677995364253,
      "learning_rate": 6.077077409299323e-06,
      "loss": 0.8136,
      "step": 1538
    },
    {
      "epoch": 0.4486553458202755,
      "grad_norm": 19.71341346374882,
      "learning_rate": 6.072467801666215e-06,
      "loss": 0.8372,
      "step": 1539
    },
    {
      "epoch": 0.4489468697616792,
      "grad_norm": 19.985339040764174,
      "learning_rate": 6.067857238342451e-06,
      "loss": 0.8447,
      "step": 1540
    },
    {
      "epoch": 0.4492383937030829,
      "grad_norm": 58.04815882749382,
      "learning_rate": 6.063245723436567e-06,
      "loss": 0.8046,
      "step": 1541
    },
    {
      "epoch": 0.4495299176444866,
      "grad_norm": 48.78216082750538,
      "learning_rate": 6.058633261057945e-06,
      "loss": 0.9348,
      "step": 1542
    },
    {
      "epoch": 0.4498214415858902,
      "grad_norm": 20.837143141537105,
      "learning_rate": 6.054019855316812e-06,
      "loss": 0.9641,
      "step": 1543
    },
    {
      "epoch": 0.4501129655272939,
      "grad_norm": 27.039700694282015,
      "learning_rate": 6.049405510324237e-06,
      "loss": 0.8443,
      "step": 1544
    },
    {
      "epoch": 0.4504044894686976,
      "grad_norm": 26.67932571451418,
      "learning_rate": 6.04479023019213e-06,
      "loss": 0.7874,
      "step": 1545
    },
    {
      "epoch": 0.4506960134101013,
      "grad_norm": 24.59000681197629,
      "learning_rate": 6.040174019033226e-06,
      "loss": 0.8928,
      "step": 1546
    },
    {
      "epoch": 0.450987537351505,
      "grad_norm": 28.019298592062086,
      "learning_rate": 6.035556880961093e-06,
      "loss": 0.9133,
      "step": 1547
    },
    {
      "epoch": 0.4512790612929087,
      "grad_norm": 32.38907888260317,
      "learning_rate": 6.030938820090128e-06,
      "loss": 0.8161,
      "step": 1548
    },
    {
      "epoch": 0.45157058523431237,
      "grad_norm": 25.519265670776754,
      "learning_rate": 6.026319840535546e-06,
      "loss": 0.9415,
      "step": 1549
    },
    {
      "epoch": 0.45186210917571606,
      "grad_norm": 36.57811872515572,
      "learning_rate": 6.021699946413384e-06,
      "loss": 0.9398,
      "step": 1550
    },
    {
      "epoch": 0.45186210917571606,
      "eval_loss": 1.1335513591766357,
      "eval_runtime": 2427.3229,
      "eval_samples_per_second": 7.115,
      "eval_steps_per_second": 0.395,
      "step": 1550
    },
    {
      "epoch": 0.45215363311711976,
      "grad_norm": 54.16440857756742,
      "learning_rate": 6.017079141840491e-06,
      "loss": 0.8535,
      "step": 1551
    },
    {
      "epoch": 0.45244515705852345,
      "grad_norm": 30.74993737621088,
      "learning_rate": 6.012457430934532e-06,
      "loss": 0.7613,
      "step": 1552
    },
    {
      "epoch": 0.45273668099992714,
      "grad_norm": 20.163080109826275,
      "learning_rate": 6.007834817813973e-06,
      "loss": 0.9442,
      "step": 1553
    },
    {
      "epoch": 0.4530282049413308,
      "grad_norm": 28.906057426868674,
      "learning_rate": 6.003211306598089e-06,
      "loss": 0.7588,
      "step": 1554
    },
    {
      "epoch": 0.4533197288827345,
      "grad_norm": 28.027964424315204,
      "learning_rate": 5.998586901406953e-06,
      "loss": 0.8609,
      "step": 1555
    },
    {
      "epoch": 0.45361125282413817,
      "grad_norm": 27.890628938355064,
      "learning_rate": 5.993961606361436e-06,
      "loss": 0.9501,
      "step": 1556
    },
    {
      "epoch": 0.45390277676554186,
      "grad_norm": 29.13782016764827,
      "learning_rate": 5.9893354255832005e-06,
      "loss": 0.7836,
      "step": 1557
    },
    {
      "epoch": 0.45419430070694555,
      "grad_norm": 20.254258216294073,
      "learning_rate": 5.984708363194702e-06,
      "loss": 0.9759,
      "step": 1558
    },
    {
      "epoch": 0.45448582464834925,
      "grad_norm": 19.469924242085877,
      "learning_rate": 5.980080423319176e-06,
      "loss": 0.8413,
      "step": 1559
    },
    {
      "epoch": 0.45477734858975294,
      "grad_norm": 17.79079273133726,
      "learning_rate": 5.975451610080643e-06,
      "loss": 0.8329,
      "step": 1560
    },
    {
      "epoch": 0.45506887253115663,
      "grad_norm": 19.041610519470346,
      "learning_rate": 5.970821927603901e-06,
      "loss": 0.721,
      "step": 1561
    },
    {
      "epoch": 0.4553603964725603,
      "grad_norm": 24.008926375452813,
      "learning_rate": 5.966191380014524e-06,
      "loss": 0.8756,
      "step": 1562
    },
    {
      "epoch": 0.455651920413964,
      "grad_norm": 46.39886994766414,
      "learning_rate": 5.961559971438858e-06,
      "loss": 0.8333,
      "step": 1563
    },
    {
      "epoch": 0.4559434443553677,
      "grad_norm": 15.746062519534892,
      "learning_rate": 5.956927706004012e-06,
      "loss": 0.7485,
      "step": 1564
    },
    {
      "epoch": 0.45623496829677135,
      "grad_norm": 25.702634624239145,
      "learning_rate": 5.95229458783786e-06,
      "loss": 0.9351,
      "step": 1565
    },
    {
      "epoch": 0.45652649223817504,
      "grad_norm": 35.98608591440873,
      "learning_rate": 5.947660621069038e-06,
      "loss": 1.0613,
      "step": 1566
    },
    {
      "epoch": 0.45681801617957873,
      "grad_norm": 18.318360437143326,
      "learning_rate": 5.9430258098269375e-06,
      "loss": 0.8418,
      "step": 1567
    },
    {
      "epoch": 0.4571095401209824,
      "grad_norm": 24.342852384013877,
      "learning_rate": 5.938390158241701e-06,
      "loss": 0.8742,
      "step": 1568
    },
    {
      "epoch": 0.4574010640623861,
      "grad_norm": 23.29905069572041,
      "learning_rate": 5.93375367044422e-06,
      "loss": 0.9555,
      "step": 1569
    },
    {
      "epoch": 0.4576925880037898,
      "grad_norm": 19.3874578312842,
      "learning_rate": 5.929116350566132e-06,
      "loss": 0.9465,
      "step": 1570
    },
    {
      "epoch": 0.4579841119451935,
      "grad_norm": 28.16823758892253,
      "learning_rate": 5.924478202739818e-06,
      "loss": 0.965,
      "step": 1571
    },
    {
      "epoch": 0.4582756358865972,
      "grad_norm": 14.721287252688436,
      "learning_rate": 5.919839231098392e-06,
      "loss": 0.6172,
      "step": 1572
    },
    {
      "epoch": 0.4585671598280009,
      "grad_norm": 18.369432689842355,
      "learning_rate": 5.915199439775706e-06,
      "loss": 0.9026,
      "step": 1573
    },
    {
      "epoch": 0.4588586837694046,
      "grad_norm": 20.71082736915286,
      "learning_rate": 5.910558832906341e-06,
      "loss": 1.0038,
      "step": 1574
    },
    {
      "epoch": 0.4591502077108082,
      "grad_norm": 20.983850690126197,
      "learning_rate": 5.9059174146256045e-06,
      "loss": 0.9068,
      "step": 1575
    },
    {
      "epoch": 0.4594417316522119,
      "grad_norm": 18.083780256807156,
      "learning_rate": 5.90127518906953e-06,
      "loss": 0.8492,
      "step": 1576
    },
    {
      "epoch": 0.4597332555936156,
      "grad_norm": 15.328154769942621,
      "learning_rate": 5.896632160374865e-06,
      "loss": 0.858,
      "step": 1577
    },
    {
      "epoch": 0.4600247795350193,
      "grad_norm": 21.34622985861086,
      "learning_rate": 5.891988332679075e-06,
      "loss": 1.0433,
      "step": 1578
    },
    {
      "epoch": 0.460316303476423,
      "grad_norm": 24.857735950833217,
      "learning_rate": 5.887343710120341e-06,
      "loss": 0.9613,
      "step": 1579
    },
    {
      "epoch": 0.4606078274178267,
      "grad_norm": 12.968725961980914,
      "learning_rate": 5.882698296837549e-06,
      "loss": 0.8398,
      "step": 1580
    },
    {
      "epoch": 0.4608993513592304,
      "grad_norm": 16.087757076031593,
      "learning_rate": 5.878052096970287e-06,
      "loss": 0.8802,
      "step": 1581
    },
    {
      "epoch": 0.4611908753006341,
      "grad_norm": 16.956393670524236,
      "learning_rate": 5.87340511465885e-06,
      "loss": 1.0957,
      "step": 1582
    },
    {
      "epoch": 0.46148239924203777,
      "grad_norm": 16.777133989820822,
      "learning_rate": 5.868757354044227e-06,
      "loss": 0.8083,
      "step": 1583
    },
    {
      "epoch": 0.46177392318344146,
      "grad_norm": 18.606417812282498,
      "learning_rate": 5.864108819268098e-06,
      "loss": 0.7525,
      "step": 1584
    },
    {
      "epoch": 0.46206544712484515,
      "grad_norm": 13.84639779003045,
      "learning_rate": 5.859459514472839e-06,
      "loss": 1.0734,
      "step": 1585
    },
    {
      "epoch": 0.4623569710662488,
      "grad_norm": 42.46800790414524,
      "learning_rate": 5.8548094438015065e-06,
      "loss": 0.8124,
      "step": 1586
    },
    {
      "epoch": 0.4626484950076525,
      "grad_norm": 20.158878284988642,
      "learning_rate": 5.850158611397844e-06,
      "loss": 0.6941,
      "step": 1587
    },
    {
      "epoch": 0.4629400189490562,
      "grad_norm": 32.30706718177147,
      "learning_rate": 5.8455070214062685e-06,
      "loss": 0.817,
      "step": 1588
    },
    {
      "epoch": 0.46323154289045987,
      "grad_norm": 21.58272154881185,
      "learning_rate": 5.840854677971878e-06,
      "loss": 0.6903,
      "step": 1589
    },
    {
      "epoch": 0.46352306683186356,
      "grad_norm": 15.121573161308367,
      "learning_rate": 5.8362015852404365e-06,
      "loss": 0.8294,
      "step": 1590
    },
    {
      "epoch": 0.46381459077326725,
      "grad_norm": 24.936882717717147,
      "learning_rate": 5.83154774735838e-06,
      "loss": 0.7823,
      "step": 1591
    },
    {
      "epoch": 0.46410611471467095,
      "grad_norm": 27.75621647853475,
      "learning_rate": 5.826893168472807e-06,
      "loss": 0.8445,
      "step": 1592
    },
    {
      "epoch": 0.46439763865607464,
      "grad_norm": 18.53224970783507,
      "learning_rate": 5.822237852731476e-06,
      "loss": 0.9756,
      "step": 1593
    },
    {
      "epoch": 0.46468916259747833,
      "grad_norm": 44.81453812607745,
      "learning_rate": 5.8175818042828e-06,
      "loss": 0.729,
      "step": 1594
    },
    {
      "epoch": 0.464980686538882,
      "grad_norm": 19.947438147909228,
      "learning_rate": 5.81292502727585e-06,
      "loss": 0.7256,
      "step": 1595
    },
    {
      "epoch": 0.4652722104802857,
      "grad_norm": 24.917586335107817,
      "learning_rate": 5.808267525860343e-06,
      "loss": 0.8935,
      "step": 1596
    },
    {
      "epoch": 0.46556373442168936,
      "grad_norm": 29.651186353529493,
      "learning_rate": 5.80360930418664e-06,
      "loss": 0.8241,
      "step": 1597
    },
    {
      "epoch": 0.46585525836309305,
      "grad_norm": 29.01177158776696,
      "learning_rate": 5.798950366405748e-06,
      "loss": 0.9988,
      "step": 1598
    },
    {
      "epoch": 0.46614678230449674,
      "grad_norm": 20.26017713706667,
      "learning_rate": 5.794290716669307e-06,
      "loss": 0.9106,
      "step": 1599
    },
    {
      "epoch": 0.46643830624590044,
      "grad_norm": 21.37686984576082,
      "learning_rate": 5.789630359129599e-06,
      "loss": 0.8258,
      "step": 1600
    },
    {
      "epoch": 0.46643830624590044,
      "eval_loss": 1.1344701051712036,
      "eval_runtime": 2381.9775,
      "eval_samples_per_second": 7.25,
      "eval_steps_per_second": 0.403,
      "step": 1600
    },
    {
      "epoch": 0.46672983018730413,
      "grad_norm": 42.16768819798762,
      "learning_rate": 5.784969297939527e-06,
      "loss": 0.8149,
      "step": 1601
    },
    {
      "epoch": 0.4670213541287078,
      "grad_norm": 18.110886339237197,
      "learning_rate": 5.780307537252629e-06,
      "loss": 0.8176,
      "step": 1602
    },
    {
      "epoch": 0.4673128780701115,
      "grad_norm": 14.853521757798603,
      "learning_rate": 5.7756450812230625e-06,
      "loss": 0.6776,
      "step": 1603
    },
    {
      "epoch": 0.4676044020115152,
      "grad_norm": 31.246130236252014,
      "learning_rate": 5.770981934005606e-06,
      "loss": 0.824,
      "step": 1604
    },
    {
      "epoch": 0.4678959259529189,
      "grad_norm": 22.162716369807114,
      "learning_rate": 5.766318099755652e-06,
      "loss": 0.7969,
      "step": 1605
    },
    {
      "epoch": 0.4681874498943226,
      "grad_norm": 25.379573594586727,
      "learning_rate": 5.76165358262921e-06,
      "loss": 0.9586,
      "step": 1606
    },
    {
      "epoch": 0.4684789738357263,
      "grad_norm": 27.345204947073668,
      "learning_rate": 5.7569883867828915e-06,
      "loss": 0.7942,
      "step": 1607
    },
    {
      "epoch": 0.4687704977771299,
      "grad_norm": 16.829754285557296,
      "learning_rate": 5.752322516373916e-06,
      "loss": 0.867,
      "step": 1608
    },
    {
      "epoch": 0.4690620217185336,
      "grad_norm": 17.66838475760037,
      "learning_rate": 5.747655975560106e-06,
      "loss": 0.844,
      "step": 1609
    },
    {
      "epoch": 0.4693535456599373,
      "grad_norm": 22.62495719088453,
      "learning_rate": 5.742988768499879e-06,
      "loss": 0.7342,
      "step": 1610
    },
    {
      "epoch": 0.469645069601341,
      "grad_norm": 39.30421111181798,
      "learning_rate": 5.738320899352246e-06,
      "loss": 0.9229,
      "step": 1611
    },
    {
      "epoch": 0.4699365935427447,
      "grad_norm": 29.530119036743585,
      "learning_rate": 5.733652372276809e-06,
      "loss": 1.0732,
      "step": 1612
    },
    {
      "epoch": 0.4702281174841484,
      "grad_norm": 21.018032314141962,
      "learning_rate": 5.7289831914337565e-06,
      "loss": 0.8229,
      "step": 1613
    },
    {
      "epoch": 0.4705196414255521,
      "grad_norm": 19.60812490963059,
      "learning_rate": 5.724313360983859e-06,
      "loss": 0.9734,
      "step": 1614
    },
    {
      "epoch": 0.4708111653669558,
      "grad_norm": 20.485526502031394,
      "learning_rate": 5.7196428850884665e-06,
      "loss": 0.7879,
      "step": 1615
    },
    {
      "epoch": 0.47110268930835947,
      "grad_norm": 25.87583040439271,
      "learning_rate": 5.7149717679095026e-06,
      "loss": 0.874,
      "step": 1616
    },
    {
      "epoch": 0.47139421324976316,
      "grad_norm": 17.14803350744817,
      "learning_rate": 5.710300013609464e-06,
      "loss": 0.8033,
      "step": 1617
    },
    {
      "epoch": 0.4716857371911668,
      "grad_norm": 16.367022845209018,
      "learning_rate": 5.705627626351415e-06,
      "loss": 0.9622,
      "step": 1618
    },
    {
      "epoch": 0.4719772611325705,
      "grad_norm": 84.21239003865936,
      "learning_rate": 5.700954610298982e-06,
      "loss": 0.8714,
      "step": 1619
    },
    {
      "epoch": 0.4722687850739742,
      "grad_norm": 26.955194015609816,
      "learning_rate": 5.6962809696163536e-06,
      "loss": 1.0274,
      "step": 1620
    },
    {
      "epoch": 0.4725603090153779,
      "grad_norm": 25.439705915060408,
      "learning_rate": 5.691606708468275e-06,
      "loss": 0.8565,
      "step": 1621
    },
    {
      "epoch": 0.47285183295678157,
      "grad_norm": 28.046564464404195,
      "learning_rate": 5.686931831020044e-06,
      "loss": 0.9042,
      "step": 1622
    },
    {
      "epoch": 0.47314335689818526,
      "grad_norm": 28.561757452178316,
      "learning_rate": 5.682256341437508e-06,
      "loss": 0.6386,
      "step": 1623
    },
    {
      "epoch": 0.47343488083958896,
      "grad_norm": 20.351210000261865,
      "learning_rate": 5.6775802438870596e-06,
      "loss": 0.6352,
      "step": 1624
    },
    {
      "epoch": 0.47372640478099265,
      "grad_norm": 64.11966872768052,
      "learning_rate": 5.672903542535631e-06,
      "loss": 0.708,
      "step": 1625
    },
    {
      "epoch": 0.47401792872239634,
      "grad_norm": 18.119943222460968,
      "learning_rate": 5.668226241550698e-06,
      "loss": 0.9858,
      "step": 1626
    },
    {
      "epoch": 0.47430945266380004,
      "grad_norm": 30.07594576594263,
      "learning_rate": 5.663548345100261e-06,
      "loss": 0.9061,
      "step": 1627
    },
    {
      "epoch": 0.47460097660520373,
      "grad_norm": 17.877192943801152,
      "learning_rate": 5.658869857352866e-06,
      "loss": 0.9248,
      "step": 1628
    },
    {
      "epoch": 0.47489250054660737,
      "grad_norm": 26.300387666092927,
      "learning_rate": 5.654190782477569e-06,
      "loss": 0.815,
      "step": 1629
    },
    {
      "epoch": 0.47518402448801106,
      "grad_norm": 15.854623523118,
      "learning_rate": 5.649511124643962e-06,
      "loss": 0.9609,
      "step": 1630
    },
    {
      "epoch": 0.47547554842941475,
      "grad_norm": 21.875419987097544,
      "learning_rate": 5.64483088802215e-06,
      "loss": 0.7863,
      "step": 1631
    },
    {
      "epoch": 0.47576707237081844,
      "grad_norm": 31.785928882447873,
      "learning_rate": 5.640150076782755e-06,
      "loss": 0.7664,
      "step": 1632
    },
    {
      "epoch": 0.47605859631222214,
      "grad_norm": 36.54474560657644,
      "learning_rate": 5.635468695096915e-06,
      "loss": 0.9663,
      "step": 1633
    },
    {
      "epoch": 0.47635012025362583,
      "grad_norm": 19.68274162471047,
      "learning_rate": 5.630786747136269e-06,
      "loss": 0.9633,
      "step": 1634
    },
    {
      "epoch": 0.4766416441950295,
      "grad_norm": 37.75790662130837,
      "learning_rate": 5.626104237072965e-06,
      "loss": 0.7059,
      "step": 1635
    },
    {
      "epoch": 0.4769331681364332,
      "grad_norm": 45.22846799341363,
      "learning_rate": 5.621421169079655e-06,
      "loss": 0.8385,
      "step": 1636
    },
    {
      "epoch": 0.4772246920778369,
      "grad_norm": 22.177300670086744,
      "learning_rate": 5.6167375473294804e-06,
      "loss": 0.7263,
      "step": 1637
    },
    {
      "epoch": 0.4775162160192406,
      "grad_norm": 62.46033805839324,
      "learning_rate": 5.612053375996082e-06,
      "loss": 0.7342,
      "step": 1638
    },
    {
      "epoch": 0.4778077399606443,
      "grad_norm": 29.920537871969188,
      "learning_rate": 5.60736865925359e-06,
      "loss": 0.848,
      "step": 1639
    },
    {
      "epoch": 0.47809926390204793,
      "grad_norm": 21.832958681069236,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.7942,
      "step": 1640
    },
    {
      "epoch": 0.4783907878434516,
      "grad_norm": 22.274785558416276,
      "learning_rate": 5.5979976062402605e-06,
      "loss": 0.92,
      "step": 1641
    },
    {
      "epoch": 0.4786823117848553,
      "grad_norm": 20.5626304861578,
      "learning_rate": 5.593311278320097e-06,
      "loss": 0.9566,
      "step": 1642
    },
    {
      "epoch": 0.478973835726259,
      "grad_norm": 33.610981903123104,
      "learning_rate": 5.588624421692177e-06,
      "loss": 0.9799,
      "step": 1643
    },
    {
      "epoch": 0.4792653596676627,
      "grad_norm": 23.725130356446247,
      "learning_rate": 5.583937040533023e-06,
      "loss": 0.8581,
      "step": 1644
    },
    {
      "epoch": 0.4795568836090664,
      "grad_norm": 35.04187270934249,
      "learning_rate": 5.5792491390196236e-06,
      "loss": 0.9104,
      "step": 1645
    },
    {
      "epoch": 0.4798484075504701,
      "grad_norm": 19.594007469290275,
      "learning_rate": 5.574560721329431e-06,
      "loss": 0.7903,
      "step": 1646
    },
    {
      "epoch": 0.4801399314918738,
      "grad_norm": 17.498990890556975,
      "learning_rate": 5.569871791640357e-06,
      "loss": 0.867,
      "step": 1647
    },
    {
      "epoch": 0.4804314554332775,
      "grad_norm": 19.64406427854032,
      "learning_rate": 5.565182354130776e-06,
      "loss": 0.8069,
      "step": 1648
    },
    {
      "epoch": 0.48072297937468117,
      "grad_norm": 31.781105146041025,
      "learning_rate": 5.5604924129795025e-06,
      "loss": 0.8126,
      "step": 1649
    },
    {
      "epoch": 0.4810145033160848,
      "grad_norm": 22.387940448491975,
      "learning_rate": 5.555801972365812e-06,
      "loss": 0.7538,
      "step": 1650
    },
    {
      "epoch": 0.4810145033160848,
      "eval_loss": 1.1298507452011108,
      "eval_runtime": 2404.737,
      "eval_samples_per_second": 7.182,
      "eval_steps_per_second": 0.399,
      "step": 1650
    },
    {
      "epoch": 0.4813060272574885,
      "grad_norm": 15.921018155551073,
      "learning_rate": 5.551111036469416e-06,
      "loss": 0.7849,
      "step": 1651
    },
    {
      "epoch": 0.4815975511988922,
      "grad_norm": 15.62766381926168,
      "learning_rate": 5.5464196094704745e-06,
      "loss": 0.8586,
      "step": 1652
    },
    {
      "epoch": 0.4818890751402959,
      "grad_norm": 51.50140415565635,
      "learning_rate": 5.54172769554958e-06,
      "loss": 0.8691,
      "step": 1653
    },
    {
      "epoch": 0.4821805990816996,
      "grad_norm": 18.779871029334622,
      "learning_rate": 5.537035298887764e-06,
      "loss": 0.8497,
      "step": 1654
    },
    {
      "epoch": 0.48247212302310327,
      "grad_norm": 15.436082513703072,
      "learning_rate": 5.532342423666482e-06,
      "loss": 0.8622,
      "step": 1655
    },
    {
      "epoch": 0.48276364696450697,
      "grad_norm": 53.10119466060774,
      "learning_rate": 5.527649074067618e-06,
      "loss": 0.8112,
      "step": 1656
    },
    {
      "epoch": 0.48305517090591066,
      "grad_norm": 29.891713341209474,
      "learning_rate": 5.522955254273482e-06,
      "loss": 0.7593,
      "step": 1657
    },
    {
      "epoch": 0.48334669484731435,
      "grad_norm": 15.858111138097765,
      "learning_rate": 5.5182609684668024e-06,
      "loss": 0.7586,
      "step": 1658
    },
    {
      "epoch": 0.48363821878871804,
      "grad_norm": 23.663968455054576,
      "learning_rate": 5.5135662208307164e-06,
      "loss": 1.0356,
      "step": 1659
    },
    {
      "epoch": 0.48392974273012174,
      "grad_norm": 17.2161370007674,
      "learning_rate": 5.508871015548781e-06,
      "loss": 0.7243,
      "step": 1660
    },
    {
      "epoch": 0.4842212666715254,
      "grad_norm": 18.034397043309212,
      "learning_rate": 5.504175356804957e-06,
      "loss": 0.8327,
      "step": 1661
    },
    {
      "epoch": 0.48451279061292907,
      "grad_norm": 37.629909923763634,
      "learning_rate": 5.49947924878361e-06,
      "loss": 0.9262,
      "step": 1662
    },
    {
      "epoch": 0.48480431455433276,
      "grad_norm": 21.751350900670875,
      "learning_rate": 5.494782695669504e-06,
      "loss": 0.8281,
      "step": 1663
    },
    {
      "epoch": 0.48509583849573645,
      "grad_norm": 85.53531558005761,
      "learning_rate": 5.490085701647805e-06,
      "loss": 0.9057,
      "step": 1664
    },
    {
      "epoch": 0.48538736243714015,
      "grad_norm": 41.357174734335764,
      "learning_rate": 5.485388270904063e-06,
      "loss": 0.8185,
      "step": 1665
    },
    {
      "epoch": 0.48567888637854384,
      "grad_norm": 15.634440500004548,
      "learning_rate": 5.480690407624227e-06,
      "loss": 0.7859,
      "step": 1666
    },
    {
      "epoch": 0.48597041031994753,
      "grad_norm": 23.85061560942979,
      "learning_rate": 5.475992115994626e-06,
      "loss": 0.8401,
      "step": 1667
    },
    {
      "epoch": 0.4862619342613512,
      "grad_norm": 38.958546131475174,
      "learning_rate": 5.47129340020197e-06,
      "loss": 0.7957,
      "step": 1668
    },
    {
      "epoch": 0.4865534582027549,
      "grad_norm": 32.22164672089303,
      "learning_rate": 5.4665942644333495e-06,
      "loss": 0.8344,
      "step": 1669
    },
    {
      "epoch": 0.4868449821441586,
      "grad_norm": 18.929157020778945,
      "learning_rate": 5.461894712876228e-06,
      "loss": 0.9578,
      "step": 1670
    },
    {
      "epoch": 0.4871365060855623,
      "grad_norm": 19.847241023525157,
      "learning_rate": 5.457194749718442e-06,
      "loss": 0.8007,
      "step": 1671
    },
    {
      "epoch": 0.48742803002696594,
      "grad_norm": 119.98965199366458,
      "learning_rate": 5.45249437914819e-06,
      "loss": 0.8158,
      "step": 1672
    },
    {
      "epoch": 0.48771955396836963,
      "grad_norm": 71.59154801328565,
      "learning_rate": 5.447793605354037e-06,
      "loss": 0.8194,
      "step": 1673
    },
    {
      "epoch": 0.4880110779097733,
      "grad_norm": 36.244949584033286,
      "learning_rate": 5.443092432524906e-06,
      "loss": 0.7408,
      "step": 1674
    },
    {
      "epoch": 0.488302601851177,
      "grad_norm": 16.11948522748229,
      "learning_rate": 5.438390864850079e-06,
      "loss": 0.9817,
      "step": 1675
    },
    {
      "epoch": 0.4885941257925807,
      "grad_norm": 19.168492686157276,
      "learning_rate": 5.433688906519183e-06,
      "loss": 0.9381,
      "step": 1676
    },
    {
      "epoch": 0.4888856497339844,
      "grad_norm": 40.33043522334688,
      "learning_rate": 5.4289865617222005e-06,
      "loss": 0.7786,
      "step": 1677
    },
    {
      "epoch": 0.4891771736753881,
      "grad_norm": 31.584661803242106,
      "learning_rate": 5.424283834649451e-06,
      "loss": 0.8338,
      "step": 1678
    },
    {
      "epoch": 0.4894686976167918,
      "grad_norm": 48.79201430887022,
      "learning_rate": 5.419580729491604e-06,
      "loss": 0.8838,
      "step": 1679
    },
    {
      "epoch": 0.4897602215581955,
      "grad_norm": 16.134471691563242,
      "learning_rate": 5.414877250439654e-06,
      "loss": 0.8521,
      "step": 1680
    },
    {
      "epoch": 0.4900517454995992,
      "grad_norm": 21.974842888858728,
      "learning_rate": 5.410173401684942e-06,
      "loss": 0.8621,
      "step": 1681
    },
    {
      "epoch": 0.4903432694410028,
      "grad_norm": 17.71842927831308,
      "learning_rate": 5.405469187419126e-06,
      "loss": 0.6764,
      "step": 1682
    },
    {
      "epoch": 0.4906347933824065,
      "grad_norm": 14.129922823328107,
      "learning_rate": 5.4007646118341975e-06,
      "loss": 0.6978,
      "step": 1683
    },
    {
      "epoch": 0.4909263173238102,
      "grad_norm": 17.131082787284804,
      "learning_rate": 5.39605967912247e-06,
      "loss": 0.8547,
      "step": 1684
    },
    {
      "epoch": 0.4912178412652139,
      "grad_norm": 18.83251473468682,
      "learning_rate": 5.3913543934765685e-06,
      "loss": 0.9994,
      "step": 1685
    },
    {
      "epoch": 0.4915093652066176,
      "grad_norm": 26.311275724380955,
      "learning_rate": 5.386648759089441e-06,
      "loss": 0.9165,
      "step": 1686
    },
    {
      "epoch": 0.4918008891480213,
      "grad_norm": 29.292462616319327,
      "learning_rate": 5.381942780154341e-06,
      "loss": 0.9128,
      "step": 1687
    },
    {
      "epoch": 0.492092413089425,
      "grad_norm": 26.90540321171476,
      "learning_rate": 5.3772364608648304e-06,
      "loss": 0.8995,
      "step": 1688
    },
    {
      "epoch": 0.49238393703082867,
      "grad_norm": 25.38001407713083,
      "learning_rate": 5.3725298054147745e-06,
      "loss": 0.9045,
      "step": 1689
    },
    {
      "epoch": 0.49267546097223236,
      "grad_norm": 25.660638095361264,
      "learning_rate": 5.367822817998338e-06,
      "loss": 0.9834,
      "step": 1690
    },
    {
      "epoch": 0.49296698491363605,
      "grad_norm": 24.78950207929352,
      "learning_rate": 5.363115502809981e-06,
      "loss": 0.8777,
      "step": 1691
    },
    {
      "epoch": 0.49325850885503975,
      "grad_norm": 12.122482125363012,
      "learning_rate": 5.358407864044456e-06,
      "loss": 0.8102,
      "step": 1692
    },
    {
      "epoch": 0.4935500327964434,
      "grad_norm": 37.07864081239141,
      "learning_rate": 5.353699905896806e-06,
      "loss": 0.9393,
      "step": 1693
    },
    {
      "epoch": 0.4938415567378471,
      "grad_norm": 39.76511061286262,
      "learning_rate": 5.348991632562355e-06,
      "loss": 0.7587,
      "step": 1694
    },
    {
      "epoch": 0.49413308067925077,
      "grad_norm": 17.561546903950777,
      "learning_rate": 5.344283048236708e-06,
      "loss": 0.8753,
      "step": 1695
    },
    {
      "epoch": 0.49442460462065446,
      "grad_norm": 20.629029218558944,
      "learning_rate": 5.339574157115752e-06,
      "loss": 0.9029,
      "step": 1696
    },
    {
      "epoch": 0.49471612856205816,
      "grad_norm": 71.99220758531412,
      "learning_rate": 5.334864963395642e-06,
      "loss": 0.9071,
      "step": 1697
    },
    {
      "epoch": 0.49500765250346185,
      "grad_norm": 29.844972016328782,
      "learning_rate": 5.330155471272804e-06,
      "loss": 0.8118,
      "step": 1698
    },
    {
      "epoch": 0.49529917644486554,
      "grad_norm": 43.08805867944815,
      "learning_rate": 5.325445684943932e-06,
      "loss": 0.9163,
      "step": 1699
    },
    {
      "epoch": 0.49559070038626923,
      "grad_norm": 18.88537260405633,
      "learning_rate": 5.320735608605979e-06,
      "loss": 0.8164,
      "step": 1700
    },
    {
      "epoch": 0.49559070038626923,
      "eval_loss": 1.1275728940963745,
      "eval_runtime": 2403.8107,
      "eval_samples_per_second": 7.184,
      "eval_steps_per_second": 0.399,
      "step": 1700
    },
    {
      "epoch": 0.4958822243276729,
      "grad_norm": 32.85812246989952,
      "learning_rate": 5.316025246456161e-06,
      "loss": 0.7607,
      "step": 1701
    },
    {
      "epoch": 0.4961737482690766,
      "grad_norm": 59.817186000810324,
      "learning_rate": 5.311314602691943e-06,
      "loss": 0.8861,
      "step": 1702
    },
    {
      "epoch": 0.4964652722104803,
      "grad_norm": 38.07856728526652,
      "learning_rate": 5.306603681511043e-06,
      "loss": 0.8377,
      "step": 1703
    },
    {
      "epoch": 0.49675679615188395,
      "grad_norm": 27.61605646316717,
      "learning_rate": 5.301892487111431e-06,
      "loss": 0.7394,
      "step": 1704
    },
    {
      "epoch": 0.49704832009328764,
      "grad_norm": 27.13589980710168,
      "learning_rate": 5.297181023691314e-06,
      "loss": 0.8188,
      "step": 1705
    },
    {
      "epoch": 0.49733984403469134,
      "grad_norm": 15.566562769297574,
      "learning_rate": 5.292469295449141e-06,
      "loss": 0.882,
      "step": 1706
    },
    {
      "epoch": 0.49763136797609503,
      "grad_norm": 91.00738128182942,
      "learning_rate": 5.287757306583598e-06,
      "loss": 1.0178,
      "step": 1707
    },
    {
      "epoch": 0.4979228919174987,
      "grad_norm": 33.082638375896984,
      "learning_rate": 5.2830450612936e-06,
      "loss": 0.931,
      "step": 1708
    },
    {
      "epoch": 0.4982144158589024,
      "grad_norm": 27.10232309012631,
      "learning_rate": 5.278332563778297e-06,
      "loss": 0.7802,
      "step": 1709
    },
    {
      "epoch": 0.4985059398003061,
      "grad_norm": 24.773298552510585,
      "learning_rate": 5.273619818237058e-06,
      "loss": 0.742,
      "step": 1710
    },
    {
      "epoch": 0.4987974637417098,
      "grad_norm": 25.29278641696399,
      "learning_rate": 5.268906828869474e-06,
      "loss": 0.8288,
      "step": 1711
    },
    {
      "epoch": 0.4990889876831135,
      "grad_norm": 35.95523411661246,
      "learning_rate": 5.264193599875353e-06,
      "loss": 0.856,
      "step": 1712
    },
    {
      "epoch": 0.4993805116245172,
      "grad_norm": 57.63585055798106,
      "learning_rate": 5.25948013545472e-06,
      "loss": 0.8842,
      "step": 1713
    },
    {
      "epoch": 0.4996720355659208,
      "grad_norm": 24.168475492811957,
      "learning_rate": 5.254766439807807e-06,
      "loss": 0.9045,
      "step": 1714
    },
    {
      "epoch": 0.4999635595073245,
      "grad_norm": 27.86035837547898,
      "learning_rate": 5.2500525171350505e-06,
      "loss": 0.7871,
      "step": 1715
    },
    {
      "epoch": 0.5002550834487283,
      "grad_norm": 33.18287204836363,
      "learning_rate": 5.245338371637091e-06,
      "loss": 0.8346,
      "step": 1716
    },
    {
      "epoch": 0.5005466073901319,
      "grad_norm": 17.702267109763792,
      "learning_rate": 5.24062400751477e-06,
      "loss": 0.7992,
      "step": 1717
    },
    {
      "epoch": 0.5008381313315357,
      "grad_norm": 20.410223162515905,
      "learning_rate": 5.235909428969119e-06,
      "loss": 0.842,
      "step": 1718
    },
    {
      "epoch": 0.5011296552729393,
      "grad_norm": 16.122362496938532,
      "learning_rate": 5.231194640201367e-06,
      "loss": 0.9008,
      "step": 1719
    },
    {
      "epoch": 0.5014211792143429,
      "grad_norm": 20.657365292059147,
      "learning_rate": 5.226479645412923e-06,
      "loss": 0.8554,
      "step": 1720
    },
    {
      "epoch": 0.5017127031557467,
      "grad_norm": 34.78453522884253,
      "learning_rate": 5.221764448805383e-06,
      "loss": 0.8628,
      "step": 1721
    },
    {
      "epoch": 0.5020042270971503,
      "grad_norm": 36.83317379999831,
      "learning_rate": 5.2170490545805255e-06,
      "loss": 0.764,
      "step": 1722
    },
    {
      "epoch": 0.5022957510385541,
      "grad_norm": 24.37781716266635,
      "learning_rate": 5.212333466940301e-06,
      "loss": 0.8212,
      "step": 1723
    },
    {
      "epoch": 0.5025872749799577,
      "grad_norm": 23.91529413518276,
      "learning_rate": 5.207617690086831e-06,
      "loss": 0.8561,
      "step": 1724
    },
    {
      "epoch": 0.5028787989213614,
      "grad_norm": 27.344470582788638,
      "learning_rate": 5.202901728222414e-06,
      "loss": 0.8339,
      "step": 1725
    },
    {
      "epoch": 0.5031703228627651,
      "grad_norm": 26.328544099725814,
      "learning_rate": 5.1981855855495035e-06,
      "loss": 0.8443,
      "step": 1726
    },
    {
      "epoch": 0.5034618468041688,
      "grad_norm": 36.705816859688,
      "learning_rate": 5.193469266270721e-06,
      "loss": 0.7073,
      "step": 1727
    },
    {
      "epoch": 0.5037533707455725,
      "grad_norm": 32.0754226407186,
      "learning_rate": 5.188752774588841e-06,
      "loss": 0.6838,
      "step": 1728
    },
    {
      "epoch": 0.5040448946869762,
      "grad_norm": 83.86420623819967,
      "learning_rate": 5.184036114706795e-06,
      "loss": 0.843,
      "step": 1729
    },
    {
      "epoch": 0.5043364186283799,
      "grad_norm": 18.651163564038125,
      "learning_rate": 5.179319290827661e-06,
      "loss": 0.7901,
      "step": 1730
    },
    {
      "epoch": 0.5046279425697835,
      "grad_norm": 22.486623035118104,
      "learning_rate": 5.174602307154667e-06,
      "loss": 0.8151,
      "step": 1731
    },
    {
      "epoch": 0.5049194665111872,
      "grad_norm": 21.984813059412296,
      "learning_rate": 5.16988516789118e-06,
      "loss": 0.8176,
      "step": 1732
    },
    {
      "epoch": 0.5052109904525909,
      "grad_norm": 32.58807936580836,
      "learning_rate": 5.165167877240707e-06,
      "loss": 0.8651,
      "step": 1733
    },
    {
      "epoch": 0.5055025143939946,
      "grad_norm": 12.378772925870697,
      "learning_rate": 5.16045043940689e-06,
      "loss": 0.7842,
      "step": 1734
    },
    {
      "epoch": 0.5057940383353983,
      "grad_norm": 23.558784840507222,
      "learning_rate": 5.155732858593502e-06,
      "loss": 0.7056,
      "step": 1735
    },
    {
      "epoch": 0.506085562276802,
      "grad_norm": 22.228614199317203,
      "learning_rate": 5.151015139004445e-06,
      "loss": 0.811,
      "step": 1736
    },
    {
      "epoch": 0.5063770862182057,
      "grad_norm": 23.065867675894815,
      "learning_rate": 5.146297284843741e-06,
      "loss": 0.9033,
      "step": 1737
    },
    {
      "epoch": 0.5066686101596094,
      "grad_norm": 16.90724647800111,
      "learning_rate": 5.141579300315536e-06,
      "loss": 0.7464,
      "step": 1738
    },
    {
      "epoch": 0.506960134101013,
      "grad_norm": 18.24311249298659,
      "learning_rate": 5.136861189624088e-06,
      "loss": 0.8363,
      "step": 1739
    },
    {
      "epoch": 0.5072516580424168,
      "grad_norm": 27.609165119822652,
      "learning_rate": 5.132142956973773e-06,
      "loss": 0.8961,
      "step": 1740
    },
    {
      "epoch": 0.5075431819838204,
      "grad_norm": 23.093290412211843,
      "learning_rate": 5.127424606569068e-06,
      "loss": 0.8995,
      "step": 1741
    },
    {
      "epoch": 0.5078347059252241,
      "grad_norm": 24.17350727504375,
      "learning_rate": 5.122706142614562e-06,
      "loss": 0.8988,
      "step": 1742
    },
    {
      "epoch": 0.5081262298666278,
      "grad_norm": 38.41877136308196,
      "learning_rate": 5.117987569314941e-06,
      "loss": 0.7469,
      "step": 1743
    },
    {
      "epoch": 0.5084177538080314,
      "grad_norm": 16.26484465954926,
      "learning_rate": 5.113268890874994e-06,
      "loss": 0.8795,
      "step": 1744
    },
    {
      "epoch": 0.5087092777494352,
      "grad_norm": 16.874600907668018,
      "learning_rate": 5.108550111499594e-06,
      "loss": 0.7949,
      "step": 1745
    },
    {
      "epoch": 0.5090008016908388,
      "grad_norm": 57.94963650579031,
      "learning_rate": 5.103831235393714e-06,
      "loss": 0.9118,
      "step": 1746
    },
    {
      "epoch": 0.5092923256322426,
      "grad_norm": 15.525234898520274,
      "learning_rate": 5.099112266762405e-06,
      "loss": 0.7778,
      "step": 1747
    },
    {
      "epoch": 0.5095838495736462,
      "grad_norm": 16.73369540803989,
      "learning_rate": 5.094393209810806e-06,
      "loss": 0.8652,
      "step": 1748
    },
    {
      "epoch": 0.50987537351505,
      "grad_norm": 32.03317790561977,
      "learning_rate": 5.089674068744134e-06,
      "loss": 0.7222,
      "step": 1749
    },
    {
      "epoch": 0.5101668974564536,
      "grad_norm": 16.639853391667163,
      "learning_rate": 5.084954847767677e-06,
      "loss": 0.8203,
      "step": 1750
    },
    {
      "epoch": 0.5101668974564536,
      "eval_loss": 1.1294647455215454,
      "eval_runtime": 2389.0657,
      "eval_samples_per_second": 7.229,
      "eval_steps_per_second": 0.402,
      "step": 1750
    },
    {
      "epoch": 0.5104584213978574,
      "grad_norm": 13.229638590396435,
      "learning_rate": 5.0802355510868e-06,
      "loss": 0.8209,
      "step": 1751
    },
    {
      "epoch": 0.510749945339261,
      "grad_norm": 27.98893573597712,
      "learning_rate": 5.07551618290693e-06,
      "loss": 0.83,
      "step": 1752
    },
    {
      "epoch": 0.5110414692806646,
      "grad_norm": 25.278256506579144,
      "learning_rate": 5.070796747433563e-06,
      "loss": 0.7223,
      "step": 1753
    },
    {
      "epoch": 0.5113329932220684,
      "grad_norm": 27.87865645951939,
      "learning_rate": 5.06607724887225e-06,
      "loss": 0.9107,
      "step": 1754
    },
    {
      "epoch": 0.511624517163472,
      "grad_norm": 22.538772530232553,
      "learning_rate": 5.0613576914286e-06,
      "loss": 0.6856,
      "step": 1755
    },
    {
      "epoch": 0.5119160411048758,
      "grad_norm": 21.333723586000634,
      "learning_rate": 5.056638079308277e-06,
      "loss": 0.8675,
      "step": 1756
    },
    {
      "epoch": 0.5122075650462794,
      "grad_norm": 17.28052877726658,
      "learning_rate": 5.05191841671699e-06,
      "loss": 0.8685,
      "step": 1757
    },
    {
      "epoch": 0.5124990889876831,
      "grad_norm": 153.3638506550656,
      "learning_rate": 5.047198707860496e-06,
      "loss": 0.9182,
      "step": 1758
    },
    {
      "epoch": 0.5127906129290868,
      "grad_norm": 23.12431129498269,
      "learning_rate": 5.04247895694459e-06,
      "loss": 1.0106,
      "step": 1759
    },
    {
      "epoch": 0.5130821368704905,
      "grad_norm": 25.683435657933845,
      "learning_rate": 5.037759168175109e-06,
      "loss": 0.6523,
      "step": 1760
    },
    {
      "epoch": 0.5133736608118942,
      "grad_norm": 22.90547714656471,
      "learning_rate": 5.033039345757917e-06,
      "loss": 0.8514,
      "step": 1761
    },
    {
      "epoch": 0.5136651847532978,
      "grad_norm": 25.131804594132646,
      "learning_rate": 5.028319493898916e-06,
      "loss": 0.8894,
      "step": 1762
    },
    {
      "epoch": 0.5139567086947016,
      "grad_norm": 16.115886082520667,
      "learning_rate": 5.023599616804028e-06,
      "loss": 0.9002,
      "step": 1763
    },
    {
      "epoch": 0.5142482326361052,
      "grad_norm": 29.531448030538648,
      "learning_rate": 5.018879718679199e-06,
      "loss": 0.8718,
      "step": 1764
    },
    {
      "epoch": 0.5145397565775089,
      "grad_norm": 22.493020223116023,
      "learning_rate": 5.014159803730396e-06,
      "loss": 0.9776,
      "step": 1765
    },
    {
      "epoch": 0.5148312805189126,
      "grad_norm": 46.38829689855971,
      "learning_rate": 5.009439876163601e-06,
      "loss": 0.9644,
      "step": 1766
    },
    {
      "epoch": 0.5151228044603163,
      "grad_norm": 24.399747278276944,
      "learning_rate": 5.004719940184803e-06,
      "loss": 0.8439,
      "step": 1767
    },
    {
      "epoch": 0.51541432840172,
      "grad_norm": 21.133214437512116,
      "learning_rate": 5e-06,
      "loss": 0.7855,
      "step": 1768
    },
    {
      "epoch": 0.5157058523431237,
      "grad_norm": 23.537390048527367,
      "learning_rate": 4.9952800598152e-06,
      "loss": 0.6747,
      "step": 1769
    },
    {
      "epoch": 0.5159973762845274,
      "grad_norm": 16.4377073992603,
      "learning_rate": 4.9905601238364006e-06,
      "loss": 0.9648,
      "step": 1770
    },
    {
      "epoch": 0.5162889002259311,
      "grad_norm": 25.768716357961107,
      "learning_rate": 4.985840196269605e-06,
      "loss": 1.0752,
      "step": 1771
    },
    {
      "epoch": 0.5165804241673347,
      "grad_norm": 27.702591747577216,
      "learning_rate": 4.981120281320801e-06,
      "loss": 0.6828,
      "step": 1772
    },
    {
      "epoch": 0.5168719481087384,
      "grad_norm": 48.31653623334429,
      "learning_rate": 4.9764003831959735e-06,
      "loss": 0.9013,
      "step": 1773
    },
    {
      "epoch": 0.5171634720501421,
      "grad_norm": 18.231146412620532,
      "learning_rate": 4.971680506101086e-06,
      "loss": 0.8803,
      "step": 1774
    },
    {
      "epoch": 0.5174549959915458,
      "grad_norm": 21.95790722589688,
      "learning_rate": 4.9669606542420835e-06,
      "loss": 0.8643,
      "step": 1775
    },
    {
      "epoch": 0.5177465199329495,
      "grad_norm": 21.31338120848184,
      "learning_rate": 4.9622408318248925e-06,
      "loss": 0.9497,
      "step": 1776
    },
    {
      "epoch": 0.5180380438743531,
      "grad_norm": 14.367502014137163,
      "learning_rate": 4.957521043055411e-06,
      "loss": 0.9159,
      "step": 1777
    },
    {
      "epoch": 0.5183295678157569,
      "grad_norm": 55.12017737865191,
      "learning_rate": 4.952801292139505e-06,
      "loss": 0.8415,
      "step": 1778
    },
    {
      "epoch": 0.5186210917571605,
      "grad_norm": 18.37212346586211,
      "learning_rate": 4.948081583283011e-06,
      "loss": 0.7412,
      "step": 1779
    },
    {
      "epoch": 0.5189126156985643,
      "grad_norm": 26.84866213999454,
      "learning_rate": 4.9433619206917234e-06,
      "loss": 0.748,
      "step": 1780
    },
    {
      "epoch": 0.5192041396399679,
      "grad_norm": 21.919535798347724,
      "learning_rate": 4.938642308571401e-06,
      "loss": 0.8359,
      "step": 1781
    },
    {
      "epoch": 0.5194956635813717,
      "grad_norm": 35.30450876472278,
      "learning_rate": 4.933922751127753e-06,
      "loss": 0.829,
      "step": 1782
    },
    {
      "epoch": 0.5197871875227753,
      "grad_norm": 18.77301277928609,
      "learning_rate": 4.9292032525664386e-06,
      "loss": 0.7355,
      "step": 1783
    },
    {
      "epoch": 0.5200787114641789,
      "grad_norm": 22.844976877211554,
      "learning_rate": 4.924483817093071e-06,
      "loss": 0.8883,
      "step": 1784
    },
    {
      "epoch": 0.5203702354055827,
      "grad_norm": 19.921991686189692,
      "learning_rate": 4.9197644489132e-06,
      "loss": 0.728,
      "step": 1785
    },
    {
      "epoch": 0.5206617593469863,
      "grad_norm": 54.06791696400004,
      "learning_rate": 4.915045152232324e-06,
      "loss": 0.8785,
      "step": 1786
    },
    {
      "epoch": 0.5209532832883901,
      "grad_norm": 22.351775889000795,
      "learning_rate": 4.910325931255869e-06,
      "loss": 0.9628,
      "step": 1787
    },
    {
      "epoch": 0.5212448072297937,
      "grad_norm": 25.00921457976755,
      "learning_rate": 4.9056067901891945e-06,
      "loss": 0.8088,
      "step": 1788
    },
    {
      "epoch": 0.5215363311711975,
      "grad_norm": 26.7638958218516,
      "learning_rate": 4.900887733237596e-06,
      "loss": 0.8014,
      "step": 1789
    },
    {
      "epoch": 0.5218278551126011,
      "grad_norm": 17.038904682901222,
      "learning_rate": 4.896168764606289e-06,
      "loss": 0.7929,
      "step": 1790
    },
    {
      "epoch": 0.5221193790540049,
      "grad_norm": 30.64674017898022,
      "learning_rate": 4.891449888500408e-06,
      "loss": 0.8149,
      "step": 1791
    },
    {
      "epoch": 0.5224109029954085,
      "grad_norm": 56.75412272085873,
      "learning_rate": 4.886731109125007e-06,
      "loss": 0.9356,
      "step": 1792
    },
    {
      "epoch": 0.5227024269368122,
      "grad_norm": 16.159585991687354,
      "learning_rate": 4.882012430685059e-06,
      "loss": 0.7492,
      "step": 1793
    },
    {
      "epoch": 0.5229939508782159,
      "grad_norm": 21.667398190863675,
      "learning_rate": 4.87729385738544e-06,
      "loss": 0.8051,
      "step": 1794
    },
    {
      "epoch": 0.5232854748196195,
      "grad_norm": 14.13114933508131,
      "learning_rate": 4.872575393430935e-06,
      "loss": 0.9963,
      "step": 1795
    },
    {
      "epoch": 0.5235769987610233,
      "grad_norm": 24.81071115625335,
      "learning_rate": 4.867857043026229e-06,
      "loss": 0.7622,
      "step": 1796
    },
    {
      "epoch": 0.5238685227024269,
      "grad_norm": 29.535227744951353,
      "learning_rate": 4.863138810375913e-06,
      "loss": 0.7563,
      "step": 1797
    },
    {
      "epoch": 0.5241600466438306,
      "grad_norm": 24.0032478593292,
      "learning_rate": 4.858420699684464e-06,
      "loss": 0.8038,
      "step": 1798
    },
    {
      "epoch": 0.5244515705852343,
      "grad_norm": 23.298120626299134,
      "learning_rate": 4.85370271515626e-06,
      "loss": 0.7128,
      "step": 1799
    },
    {
      "epoch": 0.524743094526638,
      "grad_norm": 53.2248864076682,
      "learning_rate": 4.848984860995557e-06,
      "loss": 0.7379,
      "step": 1800
    },
    {
      "epoch": 0.524743094526638,
      "eval_loss": 1.1320241689682007,
      "eval_runtime": 2412.801,
      "eval_samples_per_second": 7.158,
      "eval_steps_per_second": 0.398,
      "step": 1800
    },
    {
      "epoch": 0.5250346184680417,
      "grad_norm": 31.573311701387507,
      "learning_rate": 4.8442671414064975e-06,
      "loss": 0.8299,
      "step": 1801
    },
    {
      "epoch": 0.5253261424094454,
      "grad_norm": 24.318000657135464,
      "learning_rate": 4.839549560593111e-06,
      "loss": 0.7476,
      "step": 1802
    },
    {
      "epoch": 0.525617666350849,
      "grad_norm": 12.925545159237442,
      "learning_rate": 4.834832122759296e-06,
      "loss": 0.7945,
      "step": 1803
    },
    {
      "epoch": 0.5259091902922528,
      "grad_norm": 38.60578568014243,
      "learning_rate": 4.830114832108822e-06,
      "loss": 0.7691,
      "step": 1804
    },
    {
      "epoch": 0.5262007142336564,
      "grad_norm": 20.680728483331087,
      "learning_rate": 4.825397692845335e-06,
      "loss": 0.8148,
      "step": 1805
    },
    {
      "epoch": 0.5264922381750601,
      "grad_norm": 25.475400711431085,
      "learning_rate": 4.82068070917234e-06,
      "loss": 0.7457,
      "step": 1806
    },
    {
      "epoch": 0.5267837621164638,
      "grad_norm": 72.55328963613648,
      "learning_rate": 4.815963885293206e-06,
      "loss": 0.9053,
      "step": 1807
    },
    {
      "epoch": 0.5270752860578675,
      "grad_norm": 23.858345197981162,
      "learning_rate": 4.81124722541116e-06,
      "loss": 0.7314,
      "step": 1808
    },
    {
      "epoch": 0.5273668099992712,
      "grad_norm": 33.28089398623479,
      "learning_rate": 4.80653073372928e-06,
      "loss": 0.9414,
      "step": 1809
    },
    {
      "epoch": 0.5276583339406749,
      "grad_norm": 20.6123204814782,
      "learning_rate": 4.801814414450498e-06,
      "loss": 0.7698,
      "step": 1810
    },
    {
      "epoch": 0.5279498578820786,
      "grad_norm": 19.233779047968024,
      "learning_rate": 4.797098271777587e-06,
      "loss": 0.7915,
      "step": 1811
    },
    {
      "epoch": 0.5282413818234822,
      "grad_norm": 20.327812136403825,
      "learning_rate": 4.7923823099131694e-06,
      "loss": 0.8584,
      "step": 1812
    },
    {
      "epoch": 0.528532905764886,
      "grad_norm": 20.478816566038702,
      "learning_rate": 4.787666533059702e-06,
      "loss": 0.879,
      "step": 1813
    },
    {
      "epoch": 0.5288244297062896,
      "grad_norm": 27.199018944441107,
      "learning_rate": 4.782950945419475e-06,
      "loss": 0.8438,
      "step": 1814
    },
    {
      "epoch": 0.5291159536476934,
      "grad_norm": 18.300710558138967,
      "learning_rate": 4.778235551194618e-06,
      "loss": 0.7435,
      "step": 1815
    },
    {
      "epoch": 0.529407477589097,
      "grad_norm": 26.942433779652397,
      "learning_rate": 4.7735203545870794e-06,
      "loss": 1.0436,
      "step": 1816
    },
    {
      "epoch": 0.5296990015305006,
      "grad_norm": 18.61951019797233,
      "learning_rate": 4.768805359798634e-06,
      "loss": 0.6187,
      "step": 1817
    },
    {
      "epoch": 0.5299905254719044,
      "grad_norm": 23.522250479352593,
      "learning_rate": 4.764090571030882e-06,
      "loss": 0.7047,
      "step": 1818
    },
    {
      "epoch": 0.530282049413308,
      "grad_norm": 23.136422962791062,
      "learning_rate": 4.75937599248523e-06,
      "loss": 0.7571,
      "step": 1819
    },
    {
      "epoch": 0.5305735733547118,
      "grad_norm": 28.26749617966366,
      "learning_rate": 4.75466162836291e-06,
      "loss": 0.8229,
      "step": 1820
    },
    {
      "epoch": 0.5308650972961154,
      "grad_norm": 21.03288581258853,
      "learning_rate": 4.749947482864953e-06,
      "loss": 0.9034,
      "step": 1821
    },
    {
      "epoch": 0.5311566212375192,
      "grad_norm": 39.0462839651644,
      "learning_rate": 4.745233560192195e-06,
      "loss": 0.7001,
      "step": 1822
    },
    {
      "epoch": 0.5314481451789228,
      "grad_norm": 32.07320327254597,
      "learning_rate": 4.740519864545281e-06,
      "loss": 1.0137,
      "step": 1823
    },
    {
      "epoch": 0.5317396691203266,
      "grad_norm": 23.428572152215754,
      "learning_rate": 4.735806400124648e-06,
      "loss": 0.7944,
      "step": 1824
    },
    {
      "epoch": 0.5320311930617302,
      "grad_norm": 22.357932826052924,
      "learning_rate": 4.731093171130529e-06,
      "loss": 0.9381,
      "step": 1825
    },
    {
      "epoch": 0.5323227170031338,
      "grad_norm": 19.785545727656473,
      "learning_rate": 4.726380181762943e-06,
      "loss": 0.9224,
      "step": 1826
    },
    {
      "epoch": 0.5326142409445376,
      "grad_norm": 27.672827481086376,
      "learning_rate": 4.721667436221703e-06,
      "loss": 0.823,
      "step": 1827
    },
    {
      "epoch": 0.5329057648859412,
      "grad_norm": 20.37953600249635,
      "learning_rate": 4.716954938706401e-06,
      "loss": 0.8345,
      "step": 1828
    },
    {
      "epoch": 0.533197288827345,
      "grad_norm": 18.549937672212756,
      "learning_rate": 4.712242693416406e-06,
      "loss": 0.8793,
      "step": 1829
    },
    {
      "epoch": 0.5334888127687486,
      "grad_norm": 24.756949807317618,
      "learning_rate": 4.707530704550861e-06,
      "loss": 0.9153,
      "step": 1830
    },
    {
      "epoch": 0.5337803367101523,
      "grad_norm": 23.206753597370334,
      "learning_rate": 4.702818976308689e-06,
      "loss": 0.856,
      "step": 1831
    },
    {
      "epoch": 0.534071860651556,
      "grad_norm": 15.485840905261503,
      "learning_rate": 4.69810751288857e-06,
      "loss": 0.7899,
      "step": 1832
    },
    {
      "epoch": 0.5343633845929597,
      "grad_norm": 31.31698031246852,
      "learning_rate": 4.693396318488958e-06,
      "loss": 0.8006,
      "step": 1833
    },
    {
      "epoch": 0.5346549085343634,
      "grad_norm": 18.616388136768872,
      "learning_rate": 4.688685397308061e-06,
      "loss": 0.8219,
      "step": 1834
    },
    {
      "epoch": 0.5349464324757671,
      "grad_norm": 43.269259081257765,
      "learning_rate": 4.683974753543841e-06,
      "loss": 0.9323,
      "step": 1835
    },
    {
      "epoch": 0.5352379564171708,
      "grad_norm": 18.570494907348273,
      "learning_rate": 4.679264391394022e-06,
      "loss": 0.9215,
      "step": 1836
    },
    {
      "epoch": 0.5355294803585744,
      "grad_norm": 16.558721863017293,
      "learning_rate": 4.67455431505607e-06,
      "loss": 0.8259,
      "step": 1837
    },
    {
      "epoch": 0.5358210042999781,
      "grad_norm": 22.694161677182453,
      "learning_rate": 4.669844528727197e-06,
      "loss": 0.8145,
      "step": 1838
    },
    {
      "epoch": 0.5361125282413818,
      "grad_norm": 30.70665224656062,
      "learning_rate": 4.66513503660436e-06,
      "loss": 0.8903,
      "step": 1839
    },
    {
      "epoch": 0.5364040521827855,
      "grad_norm": 20.77093263573786,
      "learning_rate": 4.660425842884249e-06,
      "loss": 0.8385,
      "step": 1840
    },
    {
      "epoch": 0.5366955761241892,
      "grad_norm": 35.951075566221625,
      "learning_rate": 4.655716951763293e-06,
      "loss": 0.802,
      "step": 1841
    },
    {
      "epoch": 0.5369871000655929,
      "grad_norm": 16.63550601806869,
      "learning_rate": 4.651008367437646e-06,
      "loss": 0.7764,
      "step": 1842
    },
    {
      "epoch": 0.5372786240069966,
      "grad_norm": 26.062355797901485,
      "learning_rate": 4.646300094103195e-06,
      "loss": 0.7715,
      "step": 1843
    },
    {
      "epoch": 0.5375701479484003,
      "grad_norm": 37.7029444169263,
      "learning_rate": 4.641592135955545e-06,
      "loss": 0.9332,
      "step": 1844
    },
    {
      "epoch": 0.5378616718898039,
      "grad_norm": 23.324903433883264,
      "learning_rate": 4.6368844971900196e-06,
      "loss": 0.7016,
      "step": 1845
    },
    {
      "epoch": 0.5381531958312077,
      "grad_norm": 25.756768206106052,
      "learning_rate": 4.6321771820016635e-06,
      "loss": 0.6797,
      "step": 1846
    },
    {
      "epoch": 0.5384447197726113,
      "grad_norm": 28.928344775701053,
      "learning_rate": 4.627470194585228e-06,
      "loss": 0.8897,
      "step": 1847
    },
    {
      "epoch": 0.538736243714015,
      "grad_norm": 19.101291457695723,
      "learning_rate": 4.62276353913517e-06,
      "loss": 0.8108,
      "step": 1848
    },
    {
      "epoch": 0.5390277676554187,
      "grad_norm": 15.56499504710655,
      "learning_rate": 4.6180572198456606e-06,
      "loss": 0.7109,
      "step": 1849
    },
    {
      "epoch": 0.5393192915968223,
      "grad_norm": 17.555812927726873,
      "learning_rate": 4.6133512409105595e-06,
      "loss": 0.6505,
      "step": 1850
    },
    {
      "epoch": 0.5393192915968223,
      "eval_loss": 1.1259058713912964,
      "eval_runtime": 2411.3552,
      "eval_samples_per_second": 7.162,
      "eval_steps_per_second": 0.398,
      "step": 1850
    },
    {
      "epoch": 0.5396108155382261,
      "grad_norm": 17.704099554501482,
      "learning_rate": 4.608645606523432e-06,
      "loss": 0.9945,
      "step": 1851
    },
    {
      "epoch": 0.5399023394796297,
      "grad_norm": 18.962222428075556,
      "learning_rate": 4.603940320877533e-06,
      "loss": 0.9622,
      "step": 1852
    },
    {
      "epoch": 0.5401938634210335,
      "grad_norm": 28.756083676219202,
      "learning_rate": 4.599235388165803e-06,
      "loss": 0.894,
      "step": 1853
    },
    {
      "epoch": 0.5404853873624371,
      "grad_norm": 12.34548757844432,
      "learning_rate": 4.594530812580876e-06,
      "loss": 0.707,
      "step": 1854
    },
    {
      "epoch": 0.5407769113038409,
      "grad_norm": 27.48443988372566,
      "learning_rate": 4.589826598315061e-06,
      "loss": 0.762,
      "step": 1855
    },
    {
      "epoch": 0.5410684352452445,
      "grad_norm": 13.564871245965703,
      "learning_rate": 4.585122749560347e-06,
      "loss": 0.8384,
      "step": 1856
    },
    {
      "epoch": 0.5413599591866483,
      "grad_norm": 60.53192986074349,
      "learning_rate": 4.5804192705083985e-06,
      "loss": 0.6616,
      "step": 1857
    },
    {
      "epoch": 0.5416514831280519,
      "grad_norm": 34.74659523354106,
      "learning_rate": 4.575716165350549e-06,
      "loss": 0.7191,
      "step": 1858
    },
    {
      "epoch": 0.5419430070694555,
      "grad_norm": 31.722861322504492,
      "learning_rate": 4.571013438277801e-06,
      "loss": 0.8919,
      "step": 1859
    },
    {
      "epoch": 0.5422345310108593,
      "grad_norm": 56.321198873458734,
      "learning_rate": 4.566311093480818e-06,
      "loss": 0.8154,
      "step": 1860
    },
    {
      "epoch": 0.5425260549522629,
      "grad_norm": 23.306912990187435,
      "learning_rate": 4.561609135149922e-06,
      "loss": 0.9254,
      "step": 1861
    },
    {
      "epoch": 0.5428175788936667,
      "grad_norm": 22.36801642408213,
      "learning_rate": 4.556907567475094e-06,
      "loss": 0.8807,
      "step": 1862
    },
    {
      "epoch": 0.5431091028350703,
      "grad_norm": 21.026172210014614,
      "learning_rate": 4.552206394645963e-06,
      "loss": 0.7505,
      "step": 1863
    },
    {
      "epoch": 0.543400626776474,
      "grad_norm": 26.75324635180825,
      "learning_rate": 4.547505620851812e-06,
      "loss": 0.6976,
      "step": 1864
    },
    {
      "epoch": 0.5436921507178777,
      "grad_norm": 18.804290323951683,
      "learning_rate": 4.54280525028156e-06,
      "loss": 0.8625,
      "step": 1865
    },
    {
      "epoch": 0.5439836746592814,
      "grad_norm": 21.970870886552216,
      "learning_rate": 4.538105287123772e-06,
      "loss": 0.9483,
      "step": 1866
    },
    {
      "epoch": 0.5442751986006851,
      "grad_norm": 50.58863361232879,
      "learning_rate": 4.533405735566652e-06,
      "loss": 0.9429,
      "step": 1867
    },
    {
      "epoch": 0.5445667225420888,
      "grad_norm": 22.126763682622144,
      "learning_rate": 4.528706599798033e-06,
      "loss": 0.863,
      "step": 1868
    },
    {
      "epoch": 0.5448582464834925,
      "grad_norm": 29.690930668991236,
      "learning_rate": 4.524007884005376e-06,
      "loss": 0.66,
      "step": 1869
    },
    {
      "epoch": 0.5451497704248961,
      "grad_norm": 29.217837565950862,
      "learning_rate": 4.5193095923757745e-06,
      "loss": 0.812,
      "step": 1870
    },
    {
      "epoch": 0.5454412943662998,
      "grad_norm": 35.85773568887543,
      "learning_rate": 4.514611729095938e-06,
      "loss": 0.8714,
      "step": 1871
    },
    {
      "epoch": 0.5457328183077035,
      "grad_norm": 40.94551498625759,
      "learning_rate": 4.509914298352197e-06,
      "loss": 0.8513,
      "step": 1872
    },
    {
      "epoch": 0.5460243422491072,
      "grad_norm": 46.951387130450684,
      "learning_rate": 4.505217304330497e-06,
      "loss": 0.8537,
      "step": 1873
    },
    {
      "epoch": 0.5463158661905109,
      "grad_norm": 20.745540456057988,
      "learning_rate": 4.5005207512163914e-06,
      "loss": 0.8245,
      "step": 1874
    },
    {
      "epoch": 0.5466073901319146,
      "grad_norm": 17.320357228087985,
      "learning_rate": 4.4958246431950445e-06,
      "loss": 0.7346,
      "step": 1875
    },
    {
      "epoch": 0.5468989140733183,
      "grad_norm": 44.82048022043543,
      "learning_rate": 4.491128984451219e-06,
      "loss": 0.7375,
      "step": 1876
    },
    {
      "epoch": 0.547190438014722,
      "grad_norm": 18.783562159283438,
      "learning_rate": 4.486433779169284e-06,
      "loss": 0.742,
      "step": 1877
    },
    {
      "epoch": 0.5474819619561256,
      "grad_norm": 48.795429058511594,
      "learning_rate": 4.481739031533201e-06,
      "loss": 0.8212,
      "step": 1878
    },
    {
      "epoch": 0.5477734858975294,
      "grad_norm": 22.57672096207851,
      "learning_rate": 4.4770447457265184e-06,
      "loss": 0.8355,
      "step": 1879
    },
    {
      "epoch": 0.548065009838933,
      "grad_norm": 26.039579358299626,
      "learning_rate": 4.472350925932384e-06,
      "loss": 0.7815,
      "step": 1880
    },
    {
      "epoch": 0.5483565337803367,
      "grad_norm": 35.5006208777595,
      "learning_rate": 4.467657576333522e-06,
      "loss": 0.7753,
      "step": 1881
    },
    {
      "epoch": 0.5486480577217404,
      "grad_norm": 60.60263063467356,
      "learning_rate": 4.462964701112237e-06,
      "loss": 0.6958,
      "step": 1882
    },
    {
      "epoch": 0.548939581663144,
      "grad_norm": 17.652387597757738,
      "learning_rate": 4.458272304450421e-06,
      "loss": 0.8734,
      "step": 1883
    },
    {
      "epoch": 0.5492311056045478,
      "grad_norm": 33.45684037365594,
      "learning_rate": 4.453580390529526e-06,
      "loss": 0.77,
      "step": 1884
    },
    {
      "epoch": 0.5495226295459514,
      "grad_norm": 36.32523937185769,
      "learning_rate": 4.448888963530585e-06,
      "loss": 0.82,
      "step": 1885
    },
    {
      "epoch": 0.5498141534873552,
      "grad_norm": 36.17377946489095,
      "learning_rate": 4.444198027634191e-06,
      "loss": 0.7805,
      "step": 1886
    },
    {
      "epoch": 0.5501056774287588,
      "grad_norm": 22.765895400417975,
      "learning_rate": 4.439507587020499e-06,
      "loss": 0.767,
      "step": 1887
    },
    {
      "epoch": 0.5503972013701626,
      "grad_norm": 37.60227326042449,
      "learning_rate": 4.434817645869226e-06,
      "loss": 0.9593,
      "step": 1888
    },
    {
      "epoch": 0.5506887253115662,
      "grad_norm": 22.733429976406445,
      "learning_rate": 4.430128208359642e-06,
      "loss": 0.8985,
      "step": 1889
    },
    {
      "epoch": 0.5509802492529698,
      "grad_norm": 37.22915071857069,
      "learning_rate": 4.42543927867057e-06,
      "loss": 0.9568,
      "step": 1890
    },
    {
      "epoch": 0.5512717731943736,
      "grad_norm": 22.40680508235212,
      "learning_rate": 4.420750860980378e-06,
      "loss": 0.924,
      "step": 1891
    },
    {
      "epoch": 0.5515632971357772,
      "grad_norm": 20.52731247830676,
      "learning_rate": 4.416062959466978e-06,
      "loss": 0.8995,
      "step": 1892
    },
    {
      "epoch": 0.551854821077181,
      "grad_norm": 26.576655134864225,
      "learning_rate": 4.411375578307824e-06,
      "loss": 0.7806,
      "step": 1893
    },
    {
      "epoch": 0.5521463450185846,
      "grad_norm": 39.37647701360445,
      "learning_rate": 4.4066887216799055e-06,
      "loss": 0.7485,
      "step": 1894
    },
    {
      "epoch": 0.5524378689599884,
      "grad_norm": 23.058880279665164,
      "learning_rate": 4.402002393759741e-06,
      "loss": 0.9077,
      "step": 1895
    },
    {
      "epoch": 0.552729392901392,
      "grad_norm": 30.042966173957627,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.8447,
      "step": 1896
    },
    {
      "epoch": 0.5530209168427958,
      "grad_norm": 30.828879943087127,
      "learning_rate": 4.392631340746411e-06,
      "loss": 0.7822,
      "step": 1897
    },
    {
      "epoch": 0.5533124407841994,
      "grad_norm": 16.57156017270087,
      "learning_rate": 4.38794662400392e-06,
      "loss": 0.8499,
      "step": 1898
    },
    {
      "epoch": 0.5536039647256031,
      "grad_norm": 29.880310979994388,
      "learning_rate": 4.383262452670523e-06,
      "loss": 0.861,
      "step": 1899
    },
    {
      "epoch": 0.5538954886670068,
      "grad_norm": 34.01361737274533,
      "learning_rate": 4.3785788309203466e-06,
      "loss": 0.8384,
      "step": 1900
    },
    {
      "epoch": 0.5538954886670068,
      "eval_loss": 1.1246861219406128,
      "eval_runtime": 2403.2164,
      "eval_samples_per_second": 7.186,
      "eval_steps_per_second": 0.399,
      "step": 1900
    },
    {
      "epoch": 0.5541870126084104,
      "grad_norm": 18.835841697172913,
      "learning_rate": 4.373895762927036e-06,
      "loss": 0.819,
      "step": 1901
    },
    {
      "epoch": 0.5544785365498142,
      "grad_norm": 24.5239599085148,
      "learning_rate": 4.369213252863733e-06,
      "loss": 0.7427,
      "step": 1902
    },
    {
      "epoch": 0.5547700604912178,
      "grad_norm": 50.58839783209179,
      "learning_rate": 4.3645313049030866e-06,
      "loss": 0.8776,
      "step": 1903
    },
    {
      "epoch": 0.5550615844326215,
      "grad_norm": 17.757354330855804,
      "learning_rate": 4.359849923217246e-06,
      "loss": 0.8055,
      "step": 1904
    },
    {
      "epoch": 0.5553531083740252,
      "grad_norm": 77.94885818489338,
      "learning_rate": 4.355169111977852e-06,
      "loss": 0.9751,
      "step": 1905
    },
    {
      "epoch": 0.5556446323154289,
      "grad_norm": 21.483744264378764,
      "learning_rate": 4.350488875356041e-06,
      "loss": 0.873,
      "step": 1906
    },
    {
      "epoch": 0.5559361562568326,
      "grad_norm": 19.587141497000736,
      "learning_rate": 4.345809217522432e-06,
      "loss": 0.5874,
      "step": 1907
    },
    {
      "epoch": 0.5562276801982363,
      "grad_norm": 29.889140601253835,
      "learning_rate": 4.341130142647136e-06,
      "loss": 0.7744,
      "step": 1908
    },
    {
      "epoch": 0.55651920413964,
      "grad_norm": 37.05969114365219,
      "learning_rate": 4.33645165489974e-06,
      "loss": 0.7363,
      "step": 1909
    },
    {
      "epoch": 0.5568107280810437,
      "grad_norm": 24.66049505583904,
      "learning_rate": 4.331773758449303e-06,
      "loss": 0.8492,
      "step": 1910
    },
    {
      "epoch": 0.5571022520224473,
      "grad_norm": 22.9549973803232,
      "learning_rate": 4.3270964574643695e-06,
      "loss": 1.0297,
      "step": 1911
    },
    {
      "epoch": 0.557393775963851,
      "grad_norm": 23.939203510371954,
      "learning_rate": 4.322419756112943e-06,
      "loss": 0.8439,
      "step": 1912
    },
    {
      "epoch": 0.5576852999052547,
      "grad_norm": 40.06039986451344,
      "learning_rate": 4.3177436585624925e-06,
      "loss": 0.8121,
      "step": 1913
    },
    {
      "epoch": 0.5579768238466584,
      "grad_norm": 22.06370498499706,
      "learning_rate": 4.313068168979957e-06,
      "loss": 0.8475,
      "step": 1914
    },
    {
      "epoch": 0.5582683477880621,
      "grad_norm": 24.73905882634464,
      "learning_rate": 4.308393291531725e-06,
      "loss": 0.8516,
      "step": 1915
    },
    {
      "epoch": 0.5585598717294658,
      "grad_norm": 25.831657058528926,
      "learning_rate": 4.303719030383648e-06,
      "loss": 0.7329,
      "step": 1916
    },
    {
      "epoch": 0.5588513956708695,
      "grad_norm": 24.325203145763236,
      "learning_rate": 4.299045389701021e-06,
      "loss": 0.8156,
      "step": 1917
    },
    {
      "epoch": 0.5591429196122731,
      "grad_norm": 31.342509156207235,
      "learning_rate": 4.294372373648587e-06,
      "loss": 0.7811,
      "step": 1918
    },
    {
      "epoch": 0.5594344435536769,
      "grad_norm": 38.79940972437801,
      "learning_rate": 4.289699986390537e-06,
      "loss": 0.8141,
      "step": 1919
    },
    {
      "epoch": 0.5597259674950805,
      "grad_norm": 23.608485220320286,
      "learning_rate": 4.285028232090499e-06,
      "loss": 0.7509,
      "step": 1920
    },
    {
      "epoch": 0.5600174914364843,
      "grad_norm": 27.10227302154118,
      "learning_rate": 4.280357114911535e-06,
      "loss": 0.9103,
      "step": 1921
    },
    {
      "epoch": 0.5603090153778879,
      "grad_norm": 19.95900560929472,
      "learning_rate": 4.275686639016142e-06,
      "loss": 0.8353,
      "step": 1922
    },
    {
      "epoch": 0.5606005393192915,
      "grad_norm": 22.057117096544573,
      "learning_rate": 4.2710168085662435e-06,
      "loss": 0.8192,
      "step": 1923
    },
    {
      "epoch": 0.5608920632606953,
      "grad_norm": 23.47836151955238,
      "learning_rate": 4.266347627723192e-06,
      "loss": 0.8819,
      "step": 1924
    },
    {
      "epoch": 0.5611835872020989,
      "grad_norm": 20.040006613718,
      "learning_rate": 4.261679100647757e-06,
      "loss": 0.5997,
      "step": 1925
    },
    {
      "epoch": 0.5614751111435027,
      "grad_norm": 21.045290622315143,
      "learning_rate": 4.257011231500122e-06,
      "loss": 0.7225,
      "step": 1926
    },
    {
      "epoch": 0.5617666350849063,
      "grad_norm": 22.799409679381114,
      "learning_rate": 4.252344024439896e-06,
      "loss": 0.8738,
      "step": 1927
    },
    {
      "epoch": 0.5620581590263101,
      "grad_norm": 73.8057071098955,
      "learning_rate": 4.247677483626085e-06,
      "loss": 0.9007,
      "step": 1928
    },
    {
      "epoch": 0.5623496829677137,
      "grad_norm": 22.227497928250692,
      "learning_rate": 4.243011613217111e-06,
      "loss": 0.9353,
      "step": 1929
    },
    {
      "epoch": 0.5626412069091175,
      "grad_norm": 20.411280379187843,
      "learning_rate": 4.238346417370793e-06,
      "loss": 0.8098,
      "step": 1930
    },
    {
      "epoch": 0.5629327308505211,
      "grad_norm": 31.67881972347154,
      "learning_rate": 4.233681900244348e-06,
      "loss": 0.8692,
      "step": 1931
    },
    {
      "epoch": 0.5632242547919248,
      "grad_norm": 23.20648132932385,
      "learning_rate": 4.229018065994396e-06,
      "loss": 0.7136,
      "step": 1932
    },
    {
      "epoch": 0.5635157787333285,
      "grad_norm": 54.10378653406913,
      "learning_rate": 4.224354918776939e-06,
      "loss": 0.8393,
      "step": 1933
    },
    {
      "epoch": 0.5638073026747321,
      "grad_norm": 25.61734035819579,
      "learning_rate": 4.2196924627473715e-06,
      "loss": 0.8748,
      "step": 1934
    },
    {
      "epoch": 0.5640988266161359,
      "grad_norm": 32.93001330769259,
      "learning_rate": 4.215030702060474e-06,
      "loss": 0.8828,
      "step": 1935
    },
    {
      "epoch": 0.5643903505575395,
      "grad_norm": 20.750311142978052,
      "learning_rate": 4.210369640870403e-06,
      "loss": 0.8173,
      "step": 1936
    },
    {
      "epoch": 0.5646818744989432,
      "grad_norm": 51.36372368488946,
      "learning_rate": 4.205709283330694e-06,
      "loss": 0.838,
      "step": 1937
    },
    {
      "epoch": 0.5649733984403469,
      "grad_norm": 27.236663947879194,
      "learning_rate": 4.201049633594254e-06,
      "loss": 0.8166,
      "step": 1938
    },
    {
      "epoch": 0.5652649223817506,
      "grad_norm": 34.439593055643044,
      "learning_rate": 4.196390695813361e-06,
      "loss": 0.8285,
      "step": 1939
    },
    {
      "epoch": 0.5655564463231543,
      "grad_norm": 43.11472004854706,
      "learning_rate": 4.1917324741396595e-06,
      "loss": 0.7733,
      "step": 1940
    },
    {
      "epoch": 0.565847970264558,
      "grad_norm": 26.870545607690637,
      "learning_rate": 4.18707497272415e-06,
      "loss": 0.6377,
      "step": 1941
    },
    {
      "epoch": 0.5661394942059617,
      "grad_norm": 16.130907447448333,
      "learning_rate": 4.1824181957172014e-06,
      "loss": 0.7486,
      "step": 1942
    },
    {
      "epoch": 0.5664310181473654,
      "grad_norm": 31.69507215380385,
      "learning_rate": 4.177762147268527e-06,
      "loss": 0.8886,
      "step": 1943
    },
    {
      "epoch": 0.566722542088769,
      "grad_norm": 40.44949667695264,
      "learning_rate": 4.173106831527194e-06,
      "loss": 0.8967,
      "step": 1944
    },
    {
      "epoch": 0.5670140660301727,
      "grad_norm": 17.94663816906519,
      "learning_rate": 4.1684522526416215e-06,
      "loss": 0.6509,
      "step": 1945
    },
    {
      "epoch": 0.5673055899715764,
      "grad_norm": 16.879083581130228,
      "learning_rate": 4.163798414759566e-06,
      "loss": 0.7497,
      "step": 1946
    },
    {
      "epoch": 0.5675971139129801,
      "grad_norm": 19.08846731166937,
      "learning_rate": 4.159145322028124e-06,
      "loss": 0.67,
      "step": 1947
    },
    {
      "epoch": 0.5678886378543838,
      "grad_norm": 18.017894054950613,
      "learning_rate": 4.154492978593733e-06,
      "loss": 0.7826,
      "step": 1948
    },
    {
      "epoch": 0.5681801617957875,
      "grad_norm": 25.71692732144016,
      "learning_rate": 4.149841388602157e-06,
      "loss": 0.8578,
      "step": 1949
    },
    {
      "epoch": 0.5684716857371912,
      "grad_norm": 19.50398216317834,
      "learning_rate": 4.145190556198494e-06,
      "loss": 0.8516,
      "step": 1950
    },
    {
      "epoch": 0.5684716857371912,
      "eval_loss": 1.1209782361984253,
      "eval_runtime": 2417.0405,
      "eval_samples_per_second": 7.145,
      "eval_steps_per_second": 0.397,
      "step": 1950
    },
    {
      "epoch": 0.5687632096785948,
      "grad_norm": 59.94445022259216,
      "learning_rate": 4.140540485527163e-06,
      "loss": 0.8161,
      "step": 1951
    },
    {
      "epoch": 0.5690547336199986,
      "grad_norm": 26.42426395829677,
      "learning_rate": 4.135891180731903e-06,
      "loss": 0.8054,
      "step": 1952
    },
    {
      "epoch": 0.5693462575614022,
      "grad_norm": 38.621530628807136,
      "learning_rate": 4.131242645955775e-06,
      "loss": 0.7507,
      "step": 1953
    },
    {
      "epoch": 0.569637781502806,
      "grad_norm": 34.09696750214969,
      "learning_rate": 4.1265948853411506e-06,
      "loss": 0.8127,
      "step": 1954
    },
    {
      "epoch": 0.5699293054442096,
      "grad_norm": 47.291439596121414,
      "learning_rate": 4.121947903029714e-06,
      "loss": 0.6036,
      "step": 1955
    },
    {
      "epoch": 0.5702208293856132,
      "grad_norm": 17.573951758499856,
      "learning_rate": 4.1173017031624544e-06,
      "loss": 0.752,
      "step": 1956
    },
    {
      "epoch": 0.570512353327017,
      "grad_norm": 51.00730691953269,
      "learning_rate": 4.1126562898796596e-06,
      "loss": 0.7682,
      "step": 1957
    },
    {
      "epoch": 0.5708038772684206,
      "grad_norm": 25.989429633947207,
      "learning_rate": 4.108011667320926e-06,
      "loss": 0.8232,
      "step": 1958
    },
    {
      "epoch": 0.5710954012098244,
      "grad_norm": 19.984048841359307,
      "learning_rate": 4.103367839625138e-06,
      "loss": 0.9547,
      "step": 1959
    },
    {
      "epoch": 0.571386925151228,
      "grad_norm": 16.692170258820152,
      "learning_rate": 4.098724810930472e-06,
      "loss": 0.703,
      "step": 1960
    },
    {
      "epoch": 0.5716784490926318,
      "grad_norm": 20.808332384100474,
      "learning_rate": 4.094082585374397e-06,
      "loss": 0.8095,
      "step": 1961
    },
    {
      "epoch": 0.5719699730340354,
      "grad_norm": 19.60145519515751,
      "learning_rate": 4.08944116709366e-06,
      "loss": 0.7642,
      "step": 1962
    },
    {
      "epoch": 0.5722614969754392,
      "grad_norm": 28.853270417034274,
      "learning_rate": 4.084800560224296e-06,
      "loss": 0.8564,
      "step": 1963
    },
    {
      "epoch": 0.5725530209168428,
      "grad_norm": 45.68011784966392,
      "learning_rate": 4.08016076890161e-06,
      "loss": 0.7949,
      "step": 1964
    },
    {
      "epoch": 0.5728445448582464,
      "grad_norm": 36.6935770999363,
      "learning_rate": 4.0755217972601835e-06,
      "loss": 0.7721,
      "step": 1965
    },
    {
      "epoch": 0.5731360687996502,
      "grad_norm": 15.375204548410636,
      "learning_rate": 4.0708836494338695e-06,
      "loss": 0.7951,
      "step": 1966
    },
    {
      "epoch": 0.5734275927410538,
      "grad_norm": 25.092887498452278,
      "learning_rate": 4.0662463295557815e-06,
      "loss": 0.9269,
      "step": 1967
    },
    {
      "epoch": 0.5737191166824576,
      "grad_norm": 44.60228722752695,
      "learning_rate": 4.061609841758302e-06,
      "loss": 0.6867,
      "step": 1968
    },
    {
      "epoch": 0.5740106406238612,
      "grad_norm": 16.810223246943984,
      "learning_rate": 4.056974190173063e-06,
      "loss": 0.7559,
      "step": 1969
    },
    {
      "epoch": 0.574302164565265,
      "grad_norm": 21.622015727967163,
      "learning_rate": 4.0523393789309625e-06,
      "loss": 0.763,
      "step": 1970
    },
    {
      "epoch": 0.5745936885066686,
      "grad_norm": 26.087422359231034,
      "learning_rate": 4.0477054121621415e-06,
      "loss": 0.7291,
      "step": 1971
    },
    {
      "epoch": 0.5748852124480723,
      "grad_norm": 21.313983491371538,
      "learning_rate": 4.04307229399599e-06,
      "loss": 0.8588,
      "step": 1972
    },
    {
      "epoch": 0.575176736389476,
      "grad_norm": 41.62149350466336,
      "learning_rate": 4.0384400285611425e-06,
      "loss": 0.9074,
      "step": 1973
    },
    {
      "epoch": 0.5754682603308797,
      "grad_norm": 18.53790977930319,
      "learning_rate": 4.0338086199854765e-06,
      "loss": 0.6456,
      "step": 1974
    },
    {
      "epoch": 0.5757597842722834,
      "grad_norm": 25.456202174448336,
      "learning_rate": 4.029178072396099e-06,
      "loss": 0.9744,
      "step": 1975
    },
    {
      "epoch": 0.576051308213687,
      "grad_norm": 28.950480692298264,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.8988,
      "step": 1976
    },
    {
      "epoch": 0.5763428321550907,
      "grad_norm": 32.3457697891429,
      "learning_rate": 4.019919576680827e-06,
      "loss": 0.9149,
      "step": 1977
    },
    {
      "epoch": 0.5766343560964944,
      "grad_norm": 30.09295162975484,
      "learning_rate": 4.0152916368053e-06,
      "loss": 0.8513,
      "step": 1978
    },
    {
      "epoch": 0.5769258800378981,
      "grad_norm": 23.212012932759038,
      "learning_rate": 4.0106645744168e-06,
      "loss": 0.9409,
      "step": 1979
    },
    {
      "epoch": 0.5772174039793018,
      "grad_norm": 19.149003943814343,
      "learning_rate": 4.006038393638565e-06,
      "loss": 0.8521,
      "step": 1980
    },
    {
      "epoch": 0.5775089279207055,
      "grad_norm": 50.30480320456863,
      "learning_rate": 4.001413098593049e-06,
      "loss": 0.6637,
      "step": 1981
    },
    {
      "epoch": 0.5778004518621092,
      "grad_norm": 22.446761239681948,
      "learning_rate": 3.996788693401914e-06,
      "loss": 0.8032,
      "step": 1982
    },
    {
      "epoch": 0.5780919758035129,
      "grad_norm": 23.577192055565952,
      "learning_rate": 3.9921651821860285e-06,
      "loss": 0.8548,
      "step": 1983
    },
    {
      "epoch": 0.5783834997449165,
      "grad_norm": 40.64064067853515,
      "learning_rate": 3.987542569065469e-06,
      "loss": 0.8721,
      "step": 1984
    },
    {
      "epoch": 0.5786750236863203,
      "grad_norm": 18.68606414785536,
      "learning_rate": 3.9829208581595095e-06,
      "loss": 0.7687,
      "step": 1985
    },
    {
      "epoch": 0.5789665476277239,
      "grad_norm": 26.756626578867337,
      "learning_rate": 3.978300053586617e-06,
      "loss": 0.8866,
      "step": 1986
    },
    {
      "epoch": 0.5792580715691276,
      "grad_norm": 33.332649486303154,
      "learning_rate": 3.9736801594644556e-06,
      "loss": 0.8268,
      "step": 1987
    },
    {
      "epoch": 0.5795495955105313,
      "grad_norm": 26.23939322760046,
      "learning_rate": 3.969061179909872e-06,
      "loss": 0.711,
      "step": 1988
    },
    {
      "epoch": 0.579841119451935,
      "grad_norm": 42.05118031213407,
      "learning_rate": 3.964443119038908e-06,
      "loss": 0.8695,
      "step": 1989
    },
    {
      "epoch": 0.5801326433933387,
      "grad_norm": 28.436000963299705,
      "learning_rate": 3.959825980966777e-06,
      "loss": 0.6993,
      "step": 1990
    },
    {
      "epoch": 0.5804241673347423,
      "grad_norm": 26.87530993842809,
      "learning_rate": 3.95520976980787e-06,
      "loss": 0.8917,
      "step": 1991
    },
    {
      "epoch": 0.5807156912761461,
      "grad_norm": 24.76950583475885,
      "learning_rate": 3.9505944896757635e-06,
      "loss": 0.8115,
      "step": 1992
    },
    {
      "epoch": 0.5810072152175497,
      "grad_norm": 20.123640217469774,
      "learning_rate": 3.945980144683189e-06,
      "loss": 0.6395,
      "step": 1993
    },
    {
      "epoch": 0.5812987391589535,
      "grad_norm": 31.14449124228232,
      "learning_rate": 3.941366738942058e-06,
      "loss": 0.8519,
      "step": 1994
    },
    {
      "epoch": 0.5815902631003571,
      "grad_norm": 40.104546722247484,
      "learning_rate": 3.936754276563435e-06,
      "loss": 0.7428,
      "step": 1995
    },
    {
      "epoch": 0.5818817870417609,
      "grad_norm": 29.354458076582905,
      "learning_rate": 3.932142761657549e-06,
      "loss": 0.7429,
      "step": 1996
    },
    {
      "epoch": 0.5821733109831645,
      "grad_norm": 31.02212702664378,
      "learning_rate": 3.927532198333785e-06,
      "loss": 0.7896,
      "step": 1997
    },
    {
      "epoch": 0.5824648349245681,
      "grad_norm": 19.17412162747834,
      "learning_rate": 3.922922590700679e-06,
      "loss": 0.6881,
      "step": 1998
    },
    {
      "epoch": 0.5827563588659719,
      "grad_norm": 19.210356303402005,
      "learning_rate": 3.9183139428659135e-06,
      "loss": 0.9037,
      "step": 1999
    },
    {
      "epoch": 0.5830478828073755,
      "grad_norm": 23.417481914761456,
      "learning_rate": 3.913706258936317e-06,
      "loss": 0.7415,
      "step": 2000
    },
    {
      "epoch": 0.5830478828073755,
      "eval_loss": 1.1267447471618652,
      "eval_runtime": 2405.0307,
      "eval_samples_per_second": 7.181,
      "eval_steps_per_second": 0.399,
      "step": 2000
    },
    {
      "epoch": 0.5833394067487793,
      "grad_norm": 178.95085290382823,
      "learning_rate": 3.90909954301786e-06,
      "loss": 0.8313,
      "step": 2001
    },
    {
      "epoch": 0.5836309306901829,
      "grad_norm": 19.930683245251394,
      "learning_rate": 3.904493799215652e-06,
      "loss": 0.9165,
      "step": 2002
    },
    {
      "epoch": 0.5839224546315867,
      "grad_norm": 74.77612857300785,
      "learning_rate": 3.8998890316339316e-06,
      "loss": 0.6805,
      "step": 2003
    },
    {
      "epoch": 0.5842139785729903,
      "grad_norm": 18.686844605839532,
      "learning_rate": 3.895285244376068e-06,
      "loss": 0.7105,
      "step": 2004
    },
    {
      "epoch": 0.584505502514394,
      "grad_norm": 58.37039841624713,
      "learning_rate": 3.8906824415445634e-06,
      "loss": 0.834,
      "step": 2005
    },
    {
      "epoch": 0.5847970264557977,
      "grad_norm": 37.114989526416515,
      "learning_rate": 3.886080627241034e-06,
      "loss": 0.9748,
      "step": 2006
    },
    {
      "epoch": 0.5850885503972014,
      "grad_norm": 34.21764032492353,
      "learning_rate": 3.881479805566222e-06,
      "loss": 0.8022,
      "step": 2007
    },
    {
      "epoch": 0.5853800743386051,
      "grad_norm": 32.7445518839127,
      "learning_rate": 3.876879980619982e-06,
      "loss": 0.7905,
      "step": 2008
    },
    {
      "epoch": 0.5856715982800087,
      "grad_norm": 23.023690178172956,
      "learning_rate": 3.872281156501277e-06,
      "loss": 0.7709,
      "step": 2009
    },
    {
      "epoch": 0.5859631222214124,
      "grad_norm": 16.241035706250067,
      "learning_rate": 3.8676833373081864e-06,
      "loss": 0.8961,
      "step": 2010
    },
    {
      "epoch": 0.5862546461628161,
      "grad_norm": 28.883734739271162,
      "learning_rate": 3.863086527137888e-06,
      "loss": 0.8296,
      "step": 2011
    },
    {
      "epoch": 0.5865461701042198,
      "grad_norm": 25.642925748364714,
      "learning_rate": 3.8584907300866595e-06,
      "loss": 0.776,
      "step": 2012
    },
    {
      "epoch": 0.5868376940456235,
      "grad_norm": 44.497193326942984,
      "learning_rate": 3.853895950249881e-06,
      "loss": 0.753,
      "step": 2013
    },
    {
      "epoch": 0.5871292179870272,
      "grad_norm": 56.755799006691326,
      "learning_rate": 3.8493021917220225e-06,
      "loss": 0.8041,
      "step": 2014
    },
    {
      "epoch": 0.5874207419284309,
      "grad_norm": 31.926617039051873,
      "learning_rate": 3.844709458596645e-06,
      "loss": 0.8618,
      "step": 2015
    },
    {
      "epoch": 0.5877122658698346,
      "grad_norm": 38.198669650241435,
      "learning_rate": 3.840117754966396e-06,
      "loss": 0.6901,
      "step": 2016
    },
    {
      "epoch": 0.5880037898112382,
      "grad_norm": 31.861990160699065,
      "learning_rate": 3.835527084923003e-06,
      "loss": 0.73,
      "step": 2017
    },
    {
      "epoch": 0.588295313752642,
      "grad_norm": 25.12184416737642,
      "learning_rate": 3.8309374525572765e-06,
      "loss": 0.959,
      "step": 2018
    },
    {
      "epoch": 0.5885868376940456,
      "grad_norm": 43.21163266055583,
      "learning_rate": 3.826348861959099e-06,
      "loss": 0.8279,
      "step": 2019
    },
    {
      "epoch": 0.5888783616354493,
      "grad_norm": 39.40973345536332,
      "learning_rate": 3.821761317217428e-06,
      "loss": 0.6756,
      "step": 2020
    },
    {
      "epoch": 0.589169885576853,
      "grad_norm": 21.374971372784128,
      "learning_rate": 3.817174822420287e-06,
      "loss": 0.7583,
      "step": 2021
    },
    {
      "epoch": 0.5894614095182567,
      "grad_norm": 25.588777648644097,
      "learning_rate": 3.81258938165476e-06,
      "loss": 0.7529,
      "step": 2022
    },
    {
      "epoch": 0.5897529334596604,
      "grad_norm": 28.62432664581942,
      "learning_rate": 3.8080049990070005e-06,
      "loss": 0.8006,
      "step": 2023
    },
    {
      "epoch": 0.590044457401064,
      "grad_norm": 25.848936644698966,
      "learning_rate": 3.803421678562213e-06,
      "loss": 0.7813,
      "step": 2024
    },
    {
      "epoch": 0.5903359813424678,
      "grad_norm": 26.68484334810841,
      "learning_rate": 3.7988394244046533e-06,
      "loss": 0.9062,
      "step": 2025
    },
    {
      "epoch": 0.5906275052838714,
      "grad_norm": 41.07884507840121,
      "learning_rate": 3.794258240617636e-06,
      "loss": 0.768,
      "step": 2026
    },
    {
      "epoch": 0.5909190292252752,
      "grad_norm": 26.133407200213593,
      "learning_rate": 3.7896781312835107e-06,
      "loss": 0.8164,
      "step": 2027
    },
    {
      "epoch": 0.5912105531666788,
      "grad_norm": 70.609004445694,
      "learning_rate": 3.7850991004836813e-06,
      "loss": 0.8875,
      "step": 2028
    },
    {
      "epoch": 0.5915020771080824,
      "grad_norm": 20.20765708575335,
      "learning_rate": 3.78052115229858e-06,
      "loss": 0.7341,
      "step": 2029
    },
    {
      "epoch": 0.5917936010494862,
      "grad_norm": 22.04271420816328,
      "learning_rate": 3.7759442908076786e-06,
      "loss": 0.7742,
      "step": 2030
    },
    {
      "epoch": 0.5920851249908898,
      "grad_norm": 33.34053768788192,
      "learning_rate": 3.771368520089484e-06,
      "loss": 0.817,
      "step": 2031
    },
    {
      "epoch": 0.5923766489322936,
      "grad_norm": 26.20341061125398,
      "learning_rate": 3.7667938442215247e-06,
      "loss": 0.7775,
      "step": 2032
    },
    {
      "epoch": 0.5926681728736972,
      "grad_norm": 45.36800734471245,
      "learning_rate": 3.7622202672803587e-06,
      "loss": 0.8672,
      "step": 2033
    },
    {
      "epoch": 0.592959696815101,
      "grad_norm": 21.404365543551325,
      "learning_rate": 3.7576477933415612e-06,
      "loss": 0.7901,
      "step": 2034
    },
    {
      "epoch": 0.5932512207565046,
      "grad_norm": 25.03297435812175,
      "learning_rate": 3.753076426479726e-06,
      "loss": 0.921,
      "step": 2035
    },
    {
      "epoch": 0.5935427446979084,
      "grad_norm": 75.5256558290186,
      "learning_rate": 3.748506170768462e-06,
      "loss": 0.8232,
      "step": 2036
    },
    {
      "epoch": 0.593834268639312,
      "grad_norm": 21.79540172811165,
      "learning_rate": 3.743937030280386e-06,
      "loss": 0.757,
      "step": 2037
    },
    {
      "epoch": 0.5941257925807157,
      "grad_norm": 22.827914652310444,
      "learning_rate": 3.739369009087117e-06,
      "loss": 0.7521,
      "step": 2038
    },
    {
      "epoch": 0.5944173165221194,
      "grad_norm": 35.94100929562173,
      "learning_rate": 3.734802111259287e-06,
      "loss": 0.6868,
      "step": 2039
    },
    {
      "epoch": 0.594708840463523,
      "grad_norm": 35.31098173645149,
      "learning_rate": 3.7302363408665155e-06,
      "loss": 0.904,
      "step": 2040
    },
    {
      "epoch": 0.5950003644049268,
      "grad_norm": 25.87112546651023,
      "learning_rate": 3.725671701977428e-06,
      "loss": 0.7789,
      "step": 2041
    },
    {
      "epoch": 0.5952918883463304,
      "grad_norm": 22.399107309941556,
      "learning_rate": 3.721108198659633e-06,
      "loss": 0.7675,
      "step": 2042
    },
    {
      "epoch": 0.5955834122877341,
      "grad_norm": 25.690760282434372,
      "learning_rate": 3.716545834979729e-06,
      "loss": 0.6697,
      "step": 2043
    },
    {
      "epoch": 0.5958749362291378,
      "grad_norm": 31.748095468371954,
      "learning_rate": 3.7119846150033047e-06,
      "loss": 0.7546,
      "step": 2044
    },
    {
      "epoch": 0.5961664601705415,
      "grad_norm": 14.253165985364552,
      "learning_rate": 3.707424542794922e-06,
      "loss": 0.6928,
      "step": 2045
    },
    {
      "epoch": 0.5964579841119452,
      "grad_norm": 20.420920786173554,
      "learning_rate": 3.702865622418125e-06,
      "loss": 0.8512,
      "step": 2046
    },
    {
      "epoch": 0.5967495080533489,
      "grad_norm": 20.42887668995418,
      "learning_rate": 3.698307857935429e-06,
      "loss": 0.7342,
      "step": 2047
    },
    {
      "epoch": 0.5970410319947526,
      "grad_norm": 39.4996513970185,
      "learning_rate": 3.693751253408319e-06,
      "loss": 0.8515,
      "step": 2048
    },
    {
      "epoch": 0.5973325559361563,
      "grad_norm": 26.922758770132045,
      "learning_rate": 3.68919581289725e-06,
      "loss": 0.7771,
      "step": 2049
    },
    {
      "epoch": 0.5976240798775599,
      "grad_norm": 48.526719993789435,
      "learning_rate": 3.6846415404616344e-06,
      "loss": 0.6666,
      "step": 2050
    },
    {
      "epoch": 0.5976240798775599,
      "eval_loss": 1.1219985485076904,
      "eval_runtime": 2414.1374,
      "eval_samples_per_second": 7.154,
      "eval_steps_per_second": 0.398,
      "step": 2050
    },
    {
      "epoch": 0.5979156038189636,
      "grad_norm": 72.4900585337916,
      "learning_rate": 3.6800884401598463e-06,
      "loss": 0.7285,
      "step": 2051
    },
    {
      "epoch": 0.5982071277603673,
      "grad_norm": 21.037380227122966,
      "learning_rate": 3.6755365160492187e-06,
      "loss": 0.8092,
      "step": 2052
    },
    {
      "epoch": 0.598498651701771,
      "grad_norm": 26.64131715587461,
      "learning_rate": 3.6709857721860276e-06,
      "loss": 0.7844,
      "step": 2053
    },
    {
      "epoch": 0.5987901756431747,
      "grad_norm": 29.211467622317407,
      "learning_rate": 3.6664362126255087e-06,
      "loss": 0.9275,
      "step": 2054
    },
    {
      "epoch": 0.5990816995845784,
      "grad_norm": 29.628820119762874,
      "learning_rate": 3.661887841421834e-06,
      "loss": 0.8148,
      "step": 2055
    },
    {
      "epoch": 0.5993732235259821,
      "grad_norm": 24.61238278773741,
      "learning_rate": 3.657340662628116e-06,
      "loss": 0.815,
      "step": 2056
    },
    {
      "epoch": 0.5996647474673857,
      "grad_norm": 26.142864044309253,
      "learning_rate": 3.652794680296413e-06,
      "loss": 0.8632,
      "step": 2057
    },
    {
      "epoch": 0.5999562714087895,
      "grad_norm": 34.88222798199104,
      "learning_rate": 3.648249898477707e-06,
      "loss": 0.9591,
      "step": 2058
    },
    {
      "epoch": 0.6002477953501931,
      "grad_norm": 38.394297789758795,
      "learning_rate": 3.643706321221919e-06,
      "loss": 0.8616,
      "step": 2059
    },
    {
      "epoch": 0.6005393192915969,
      "grad_norm": 45.149162700220394,
      "learning_rate": 3.6391639525778915e-06,
      "loss": 0.7433,
      "step": 2060
    },
    {
      "epoch": 0.6008308432330005,
      "grad_norm": 68.45418263336542,
      "learning_rate": 3.6346227965933872e-06,
      "loss": 0.7226,
      "step": 2061
    },
    {
      "epoch": 0.6011223671744041,
      "grad_norm": 67.70005683236215,
      "learning_rate": 3.6300828573150977e-06,
      "loss": 0.8017,
      "step": 2062
    },
    {
      "epoch": 0.6014138911158079,
      "grad_norm": 24.421101186996452,
      "learning_rate": 3.6255441387886216e-06,
      "loss": 0.8872,
      "step": 2063
    },
    {
      "epoch": 0.6017054150572115,
      "grad_norm": 26.083840279389864,
      "learning_rate": 3.621006645058472e-06,
      "loss": 0.8257,
      "step": 2064
    },
    {
      "epoch": 0.6019969389986153,
      "grad_norm": 36.18557392684573,
      "learning_rate": 3.6164703801680722e-06,
      "loss": 0.6399,
      "step": 2065
    },
    {
      "epoch": 0.6022884629400189,
      "grad_norm": 25.782493439210388,
      "learning_rate": 3.6119353481597504e-06,
      "loss": 0.8681,
      "step": 2066
    },
    {
      "epoch": 0.6025799868814227,
      "grad_norm": 47.01855600097079,
      "learning_rate": 3.6074015530747354e-06,
      "loss": 0.7697,
      "step": 2067
    },
    {
      "epoch": 0.6028715108228263,
      "grad_norm": 32.0697864973581,
      "learning_rate": 3.6028689989531533e-06,
      "loss": 0.7038,
      "step": 2068
    },
    {
      "epoch": 0.60316303476423,
      "grad_norm": 25.850109416226918,
      "learning_rate": 3.5983376898340227e-06,
      "loss": 0.6297,
      "step": 2069
    },
    {
      "epoch": 0.6034545587056337,
      "grad_norm": 22.109299704115802,
      "learning_rate": 3.593807629755258e-06,
      "loss": 0.7124,
      "step": 2070
    },
    {
      "epoch": 0.6037460826470374,
      "grad_norm": 75.04161179242074,
      "learning_rate": 3.5892788227536557e-06,
      "loss": 0.7958,
      "step": 2071
    },
    {
      "epoch": 0.6040376065884411,
      "grad_norm": 23.378263442731278,
      "learning_rate": 3.584751272864899e-06,
      "loss": 0.7873,
      "step": 2072
    },
    {
      "epoch": 0.6043291305298447,
      "grad_norm": 53.02186469423562,
      "learning_rate": 3.5802249841235474e-06,
      "loss": 0.9767,
      "step": 2073
    },
    {
      "epoch": 0.6046206544712485,
      "grad_norm": 44.62436478933056,
      "learning_rate": 3.575699960563038e-06,
      "loss": 0.7907,
      "step": 2074
    },
    {
      "epoch": 0.6049121784126521,
      "grad_norm": 34.52826014084687,
      "learning_rate": 3.571176206215683e-06,
      "loss": 1.0006,
      "step": 2075
    },
    {
      "epoch": 0.6052037023540559,
      "grad_norm": 17.740721439007995,
      "learning_rate": 3.566653725112661e-06,
      "loss": 0.8475,
      "step": 2076
    },
    {
      "epoch": 0.6054952262954595,
      "grad_norm": 29.08966971193385,
      "learning_rate": 3.5621325212840143e-06,
      "loss": 0.7453,
      "step": 2077
    },
    {
      "epoch": 0.6057867502368632,
      "grad_norm": 18.195421190432395,
      "learning_rate": 3.557612598758652e-06,
      "loss": 0.8568,
      "step": 2078
    },
    {
      "epoch": 0.6060782741782669,
      "grad_norm": 34.088831941824175,
      "learning_rate": 3.553093961564337e-06,
      "loss": 0.7966,
      "step": 2079
    },
    {
      "epoch": 0.6063697981196706,
      "grad_norm": 26.62480869582223,
      "learning_rate": 3.5485766137276894e-06,
      "loss": 0.7661,
      "step": 2080
    },
    {
      "epoch": 0.6066613220610743,
      "grad_norm": 22.72646787140385,
      "learning_rate": 3.5440605592741776e-06,
      "loss": 0.7612,
      "step": 2081
    },
    {
      "epoch": 0.606952846002478,
      "grad_norm": 28.047702439822295,
      "learning_rate": 3.5395458022281205e-06,
      "loss": 0.8242,
      "step": 2082
    },
    {
      "epoch": 0.6072443699438816,
      "grad_norm": 21.73327938427742,
      "learning_rate": 3.535032346612681e-06,
      "loss": 0.7233,
      "step": 2083
    },
    {
      "epoch": 0.6075358938852853,
      "grad_norm": 32.46813765364593,
      "learning_rate": 3.5305201964498557e-06,
      "loss": 0.8992,
      "step": 2084
    },
    {
      "epoch": 0.607827417826689,
      "grad_norm": 57.38170970588399,
      "learning_rate": 3.5260093557604903e-06,
      "loss": 0.789,
      "step": 2085
    },
    {
      "epoch": 0.6081189417680927,
      "grad_norm": 22.75574146441863,
      "learning_rate": 3.5214998285642517e-06,
      "loss": 0.7385,
      "step": 2086
    },
    {
      "epoch": 0.6084104657094964,
      "grad_norm": 25.62600346623999,
      "learning_rate": 3.5169916188796397e-06,
      "loss": 0.781,
      "step": 2087
    },
    {
      "epoch": 0.6087019896509,
      "grad_norm": 24.29654515190406,
      "learning_rate": 3.5124847307239863e-06,
      "loss": 0.7468,
      "step": 2088
    },
    {
      "epoch": 0.6089935135923038,
      "grad_norm": 99.07640927945732,
      "learning_rate": 3.507979168113437e-06,
      "loss": 0.7376,
      "step": 2089
    },
    {
      "epoch": 0.6092850375337074,
      "grad_norm": 42.174405381383025,
      "learning_rate": 3.5034749350629593e-06,
      "loss": 0.8116,
      "step": 2090
    },
    {
      "epoch": 0.6095765614751112,
      "grad_norm": 21.600213479479713,
      "learning_rate": 3.498972035586341e-06,
      "loss": 0.6901,
      "step": 2091
    },
    {
      "epoch": 0.6098680854165148,
      "grad_norm": 27.074990644648658,
      "learning_rate": 3.4944704736961722e-06,
      "loss": 0.7462,
      "step": 2092
    },
    {
      "epoch": 0.6101596093579185,
      "grad_norm": 23.96228772609966,
      "learning_rate": 3.48997025340386e-06,
      "loss": 0.6402,
      "step": 2093
    },
    {
      "epoch": 0.6104511332993222,
      "grad_norm": 22.245216947802884,
      "learning_rate": 3.4854713787196105e-06,
      "loss": 0.8267,
      "step": 2094
    },
    {
      "epoch": 0.6107426572407259,
      "grad_norm": 29.921865580965207,
      "learning_rate": 3.4809738536524306e-06,
      "loss": 0.8254,
      "step": 2095
    },
    {
      "epoch": 0.6110341811821296,
      "grad_norm": 17.90478258994544,
      "learning_rate": 3.4764776822101275e-06,
      "loss": 0.611,
      "step": 2096
    },
    {
      "epoch": 0.6113257051235332,
      "grad_norm": 23.907182104569486,
      "learning_rate": 3.4719828683993003e-06,
      "loss": 0.6742,
      "step": 2097
    },
    {
      "epoch": 0.611617229064937,
      "grad_norm": 20.530842055221076,
      "learning_rate": 3.4674894162253404e-06,
      "loss": 0.6331,
      "step": 2098
    },
    {
      "epoch": 0.6119087530063406,
      "grad_norm": 21.275713910206086,
      "learning_rate": 3.462997329692423e-06,
      "loss": 0.7192,
      "step": 2099
    },
    {
      "epoch": 0.6122002769477444,
      "grad_norm": 40.315755901077765,
      "learning_rate": 3.458506612803505e-06,
      "loss": 0.8176,
      "step": 2100
    },
    {
      "epoch": 0.6122002769477444,
      "eval_loss": 1.125646710395813,
      "eval_runtime": 2425.282,
      "eval_samples_per_second": 7.121,
      "eval_steps_per_second": 0.396,
      "step": 2100
    },
    {
      "epoch": 0.612491800889148,
      "grad_norm": 16.280940512822504,
      "learning_rate": 3.4540172695603302e-06,
      "loss": 0.9341,
      "step": 2101
    },
    {
      "epoch": 0.6127833248305518,
      "grad_norm": 93.94974872723857,
      "learning_rate": 3.4495293039634113e-06,
      "loss": 0.7934,
      "step": 2102
    },
    {
      "epoch": 0.6130748487719554,
      "grad_norm": 22.334592747669678,
      "learning_rate": 3.445042720012033e-06,
      "loss": 0.8654,
      "step": 2103
    },
    {
      "epoch": 0.613366372713359,
      "grad_norm": 26.987116719153743,
      "learning_rate": 3.440557521704256e-06,
      "loss": 0.7093,
      "step": 2104
    },
    {
      "epoch": 0.6136578966547628,
      "grad_norm": 37.79484867396204,
      "learning_rate": 3.436073713036897e-06,
      "loss": 0.8584,
      "step": 2105
    },
    {
      "epoch": 0.6139494205961664,
      "grad_norm": 29.474823988824244,
      "learning_rate": 3.4315912980055433e-06,
      "loss": 0.8607,
      "step": 2106
    },
    {
      "epoch": 0.6142409445375702,
      "grad_norm": 16.780566572334727,
      "learning_rate": 3.4271102806045336e-06,
      "loss": 0.6792,
      "step": 2107
    },
    {
      "epoch": 0.6145324684789738,
      "grad_norm": 35.0380664396021,
      "learning_rate": 3.4226306648269616e-06,
      "loss": 0.7759,
      "step": 2108
    },
    {
      "epoch": 0.6148239924203776,
      "grad_norm": 15.274461472886774,
      "learning_rate": 3.4181524546646785e-06,
      "loss": 0.8438,
      "step": 2109
    },
    {
      "epoch": 0.6151155163617812,
      "grad_norm": 28.295619055007634,
      "learning_rate": 3.413675654108275e-06,
      "loss": 0.739,
      "step": 2110
    },
    {
      "epoch": 0.6154070403031849,
      "grad_norm": 61.9562649232314,
      "learning_rate": 3.409200267147089e-06,
      "loss": 0.7175,
      "step": 2111
    },
    {
      "epoch": 0.6156985642445886,
      "grad_norm": 31.039069498238618,
      "learning_rate": 3.4047262977692014e-06,
      "loss": 0.7586,
      "step": 2112
    },
    {
      "epoch": 0.6159900881859923,
      "grad_norm": 19.572362454209607,
      "learning_rate": 3.400253749961424e-06,
      "loss": 0.7348,
      "step": 2113
    },
    {
      "epoch": 0.616281612127396,
      "grad_norm": 22.645818954373336,
      "learning_rate": 3.3957826277093074e-06,
      "loss": 0.8329,
      "step": 2114
    },
    {
      "epoch": 0.6165731360687996,
      "grad_norm": 34.91495629773395,
      "learning_rate": 3.391312934997128e-06,
      "loss": 0.7878,
      "step": 2115
    },
    {
      "epoch": 0.6168646600102033,
      "grad_norm": 19.466575730323026,
      "learning_rate": 3.3868446758078897e-06,
      "loss": 0.8284,
      "step": 2116
    },
    {
      "epoch": 0.617156183951607,
      "grad_norm": 19.58298317430564,
      "learning_rate": 3.382377854123322e-06,
      "loss": 0.7607,
      "step": 2117
    },
    {
      "epoch": 0.6174477078930107,
      "grad_norm": 28.925260384721295,
      "learning_rate": 3.3779124739238657e-06,
      "loss": 0.8027,
      "step": 2118
    },
    {
      "epoch": 0.6177392318344144,
      "grad_norm": 17.183627452919346,
      "learning_rate": 3.373448539188686e-06,
      "loss": 0.852,
      "step": 2119
    },
    {
      "epoch": 0.6180307557758181,
      "grad_norm": 33.132157833788455,
      "learning_rate": 3.3689860538956547e-06,
      "loss": 0.662,
      "step": 2120
    },
    {
      "epoch": 0.6183222797172218,
      "grad_norm": 29.754057817181998,
      "learning_rate": 3.364525022021351e-06,
      "loss": 0.5787,
      "step": 2121
    },
    {
      "epoch": 0.6186138036586255,
      "grad_norm": 16.73165939987113,
      "learning_rate": 3.3600654475410643e-06,
      "loss": 0.7819,
      "step": 2122
    },
    {
      "epoch": 0.6189053276000291,
      "grad_norm": 80.79316949712683,
      "learning_rate": 3.3556073344287798e-06,
      "loss": 0.7695,
      "step": 2123
    },
    {
      "epoch": 0.6191968515414329,
      "grad_norm": 16.366941370782147,
      "learning_rate": 3.351150686657185e-06,
      "loss": 0.82,
      "step": 2124
    },
    {
      "epoch": 0.6194883754828365,
      "grad_norm": 45.91590620599226,
      "learning_rate": 3.346695508197659e-06,
      "loss": 0.7883,
      "step": 2125
    },
    {
      "epoch": 0.6197798994242402,
      "grad_norm": 22.361945556085093,
      "learning_rate": 3.3422418030202696e-06,
      "loss": 0.8167,
      "step": 2126
    },
    {
      "epoch": 0.6200714233656439,
      "grad_norm": 41.884558412747396,
      "learning_rate": 3.337789575093776e-06,
      "loss": 0.7719,
      "step": 2127
    },
    {
      "epoch": 0.6203629473070476,
      "grad_norm": 23.722518034674618,
      "learning_rate": 3.3333388283856195e-06,
      "loss": 0.8218,
      "step": 2128
    },
    {
      "epoch": 0.6206544712484513,
      "grad_norm": 29.578638974359045,
      "learning_rate": 3.328889566861919e-06,
      "loss": 0.7286,
      "step": 2129
    },
    {
      "epoch": 0.6209459951898549,
      "grad_norm": 23.66871757281067,
      "learning_rate": 3.324441794487475e-06,
      "loss": 0.7079,
      "step": 2130
    },
    {
      "epoch": 0.6212375191312587,
      "grad_norm": 25.731223955503356,
      "learning_rate": 3.319995515225754e-06,
      "loss": 0.7954,
      "step": 2131
    },
    {
      "epoch": 0.6215290430726623,
      "grad_norm": 22.192322843602774,
      "learning_rate": 3.3155507330389004e-06,
      "loss": 0.6765,
      "step": 2132
    },
    {
      "epoch": 0.6218205670140661,
      "grad_norm": 32.38566820607962,
      "learning_rate": 3.3111074518877185e-06,
      "loss": 0.7242,
      "step": 2133
    },
    {
      "epoch": 0.6221120909554697,
      "grad_norm": 29.22284016975251,
      "learning_rate": 3.306665675731674e-06,
      "loss": 0.646,
      "step": 2134
    },
    {
      "epoch": 0.6224036148968735,
      "grad_norm": 20.82606914508671,
      "learning_rate": 3.302225408528898e-06,
      "loss": 0.702,
      "step": 2135
    },
    {
      "epoch": 0.6226951388382771,
      "grad_norm": 31.51751444254251,
      "learning_rate": 3.297786654236169e-06,
      "loss": 0.8024,
      "step": 2136
    },
    {
      "epoch": 0.6229866627796807,
      "grad_norm": 16.31538185016397,
      "learning_rate": 3.293349416808926e-06,
      "loss": 0.5789,
      "step": 2137
    },
    {
      "epoch": 0.6232781867210845,
      "grad_norm": 33.97338658704836,
      "learning_rate": 3.28891370020125e-06,
      "loss": 0.7753,
      "step": 2138
    },
    {
      "epoch": 0.6235697106624881,
      "grad_norm": 53.49206226114068,
      "learning_rate": 3.2844795083658656e-06,
      "loss": 0.8363,
      "step": 2139
    },
    {
      "epoch": 0.6238612346038919,
      "grad_norm": 265.4467725798008,
      "learning_rate": 3.280046845254145e-06,
      "loss": 0.8208,
      "step": 2140
    },
    {
      "epoch": 0.6241527585452955,
      "grad_norm": 19.38191824711622,
      "learning_rate": 3.275615714816094e-06,
      "loss": 0.7887,
      "step": 2141
    },
    {
      "epoch": 0.6244442824866993,
      "grad_norm": 33.618400441300594,
      "learning_rate": 3.2711861210003503e-06,
      "loss": 0.6754,
      "step": 2142
    },
    {
      "epoch": 0.6247358064281029,
      "grad_norm": 35.23860298694567,
      "learning_rate": 3.26675806775419e-06,
      "loss": 0.8293,
      "step": 2143
    },
    {
      "epoch": 0.6250273303695066,
      "grad_norm": 21.069346616643973,
      "learning_rate": 3.2623315590235076e-06,
      "loss": 0.691,
      "step": 2144
    },
    {
      "epoch": 0.6253188543109103,
      "grad_norm": 28.026385237090153,
      "learning_rate": 3.257906598752828e-06,
      "loss": 0.754,
      "step": 2145
    },
    {
      "epoch": 0.625610378252314,
      "grad_norm": 25.458732355625735,
      "learning_rate": 3.2534831908852914e-06,
      "loss": 0.8539,
      "step": 2146
    },
    {
      "epoch": 0.6259019021937177,
      "grad_norm": 33.899931565343394,
      "learning_rate": 3.249061339362658e-06,
      "loss": 0.7839,
      "step": 2147
    },
    {
      "epoch": 0.6261934261351213,
      "grad_norm": 25.487303764204665,
      "learning_rate": 3.244641048125301e-06,
      "loss": 0.7878,
      "step": 2148
    },
    {
      "epoch": 0.626484950076525,
      "grad_norm": 26.428469458966717,
      "learning_rate": 3.240222321112198e-06,
      "loss": 0.6914,
      "step": 2149
    },
    {
      "epoch": 0.6267764740179287,
      "grad_norm": 34.502315240670114,
      "learning_rate": 3.235805162260942e-06,
      "loss": 0.876,
      "step": 2150
    },
    {
      "epoch": 0.6267764740179287,
      "eval_loss": 1.1186778545379639,
      "eval_runtime": 2271.399,
      "eval_samples_per_second": 7.603,
      "eval_steps_per_second": 0.423,
      "step": 2150
    },
    {
      "epoch": 0.6270679979593324,
      "grad_norm": 80.70072320860201,
      "learning_rate": 3.2313895755077205e-06,
      "loss": 0.7069,
      "step": 2151
    },
    {
      "epoch": 0.6273595219007361,
      "grad_norm": 32.2326513687757,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.7377,
      "step": 2152
    },
    {
      "epoch": 0.6276510458421398,
      "grad_norm": 42.35175816584448,
      "learning_rate": 3.2225631340331344e-06,
      "loss": 0.8416,
      "step": 2153
    },
    {
      "epoch": 0.6279425697835435,
      "grad_norm": 45.60488888498005,
      "learning_rate": 3.218152287177133e-06,
      "loss": 0.6543,
      "step": 2154
    },
    {
      "epoch": 0.6282340937249472,
      "grad_norm": 28.28743492598315,
      "learning_rate": 3.2137430281498825e-06,
      "loss": 0.7691,
      "step": 2155
    },
    {
      "epoch": 0.6285256176663508,
      "grad_norm": 18.724740471606662,
      "learning_rate": 3.2093353608805368e-06,
      "loss": 0.7313,
      "step": 2156
    },
    {
      "epoch": 0.6288171416077545,
      "grad_norm": 25.412847750280736,
      "learning_rate": 3.2049292892968243e-06,
      "loss": 0.7037,
      "step": 2157
    },
    {
      "epoch": 0.6291086655491582,
      "grad_norm": 18.567150670345036,
      "learning_rate": 3.2005248173250593e-06,
      "loss": 0.8116,
      "step": 2158
    },
    {
      "epoch": 0.6294001894905619,
      "grad_norm": 17.938914147234247,
      "learning_rate": 3.196121948890125e-06,
      "loss": 0.7757,
      "step": 2159
    },
    {
      "epoch": 0.6296917134319656,
      "grad_norm": 21.53144631208358,
      "learning_rate": 3.1917206879154762e-06,
      "loss": 0.843,
      "step": 2160
    },
    {
      "epoch": 0.6299832373733693,
      "grad_norm": 23.464740793435528,
      "learning_rate": 3.1873210383231374e-06,
      "loss": 0.8381,
      "step": 2161
    },
    {
      "epoch": 0.630274761314773,
      "grad_norm": 24.863437002093207,
      "learning_rate": 3.1829230040336967e-06,
      "loss": 0.7597,
      "step": 2162
    },
    {
      "epoch": 0.6305662852561766,
      "grad_norm": 51.386739089708676,
      "learning_rate": 3.1785265889663024e-06,
      "loss": 0.7402,
      "step": 2163
    },
    {
      "epoch": 0.6308578091975804,
      "grad_norm": 25.083983866636576,
      "learning_rate": 3.1741317970386597e-06,
      "loss": 0.9119,
      "step": 2164
    },
    {
      "epoch": 0.631149333138984,
      "grad_norm": 22.255660410928414,
      "learning_rate": 3.1697386321670254e-06,
      "loss": 0.7948,
      "step": 2165
    },
    {
      "epoch": 0.6314408570803878,
      "grad_norm": 20.80833123818686,
      "learning_rate": 3.1653470982662114e-06,
      "loss": 0.7347,
      "step": 2166
    },
    {
      "epoch": 0.6317323810217914,
      "grad_norm": 35.84558736498403,
      "learning_rate": 3.1609571992495738e-06,
      "loss": 0.9672,
      "step": 2167
    },
    {
      "epoch": 0.632023904963195,
      "grad_norm": 29.529909363074953,
      "learning_rate": 3.1565689390290067e-06,
      "loss": 0.7946,
      "step": 2168
    },
    {
      "epoch": 0.6323154289045988,
      "grad_norm": 25.323579481627164,
      "learning_rate": 3.1521823215149535e-06,
      "loss": 0.7451,
      "step": 2169
    },
    {
      "epoch": 0.6326069528460024,
      "grad_norm": 39.4915175337859,
      "learning_rate": 3.147797350616385e-06,
      "loss": 0.5757,
      "step": 2170
    },
    {
      "epoch": 0.6328984767874062,
      "grad_norm": 22.873082042938254,
      "learning_rate": 3.143414030240813e-06,
      "loss": 0.7099,
      "step": 2171
    },
    {
      "epoch": 0.6331900007288098,
      "grad_norm": 18.92043709375269,
      "learning_rate": 3.139032364294271e-06,
      "loss": 0.8782,
      "step": 2172
    },
    {
      "epoch": 0.6334815246702136,
      "grad_norm": 22.57489758297193,
      "learning_rate": 3.134652356681319e-06,
      "loss": 0.9939,
      "step": 2173
    },
    {
      "epoch": 0.6337730486116172,
      "grad_norm": 25.611300549619916,
      "learning_rate": 3.130274011305047e-06,
      "loss": 0.735,
      "step": 2174
    },
    {
      "epoch": 0.634064572553021,
      "grad_norm": 90.4257462047812,
      "learning_rate": 3.125897332067053e-06,
      "loss": 0.9158,
      "step": 2175
    },
    {
      "epoch": 0.6343560964944246,
      "grad_norm": 40.75537456209329,
      "learning_rate": 3.1215223228674587e-06,
      "loss": 0.6073,
      "step": 2176
    },
    {
      "epoch": 0.6346476204358283,
      "grad_norm": 18.603513202574177,
      "learning_rate": 3.117148987604892e-06,
      "loss": 0.6947,
      "step": 2177
    },
    {
      "epoch": 0.634939144377232,
      "grad_norm": 34.17238950196663,
      "learning_rate": 3.1127773301764935e-06,
      "loss": 0.8267,
      "step": 2178
    },
    {
      "epoch": 0.6352306683186356,
      "grad_norm": 24.339435208576372,
      "learning_rate": 3.1084073544779065e-06,
      "loss": 0.7681,
      "step": 2179
    },
    {
      "epoch": 0.6355221922600394,
      "grad_norm": 14.568432891601999,
      "learning_rate": 3.1040390644032746e-06,
      "loss": 0.6166,
      "step": 2180
    },
    {
      "epoch": 0.635813716201443,
      "grad_norm": 23.48287518450522,
      "learning_rate": 3.0996724638452397e-06,
      "loss": 0.6862,
      "step": 2181
    },
    {
      "epoch": 0.6361052401428468,
      "grad_norm": 24.12308022256533,
      "learning_rate": 3.095307556694942e-06,
      "loss": 0.8063,
      "step": 2182
    },
    {
      "epoch": 0.6363967640842504,
      "grad_norm": 31.991295794493315,
      "learning_rate": 3.0909443468420063e-06,
      "loss": 0.7578,
      "step": 2183
    },
    {
      "epoch": 0.6366882880256541,
      "grad_norm": 27.354145020583154,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.8074,
      "step": 2184
    },
    {
      "epoch": 0.6369798119670578,
      "grad_norm": 36.30770256284456,
      "learning_rate": 3.0822230345791764e-06,
      "loss": 0.892,
      "step": 2185
    },
    {
      "epoch": 0.6372713359084615,
      "grad_norm": 114.69292672225818,
      "learning_rate": 3.077864939940959e-06,
      "loss": 0.7074,
      "step": 2186
    },
    {
      "epoch": 0.6375628598498652,
      "grad_norm": 30.093072845383748,
      "learning_rate": 3.0735085581434596e-06,
      "loss": 0.9744,
      "step": 2187
    },
    {
      "epoch": 0.6378543837912689,
      "grad_norm": 36.10656892464396,
      "learning_rate": 3.0691538930687076e-06,
      "loss": 0.7588,
      "step": 2188
    },
    {
      "epoch": 0.6381459077326725,
      "grad_norm": 47.824537704942394,
      "learning_rate": 3.0648009485972064e-06,
      "loss": 0.7609,
      "step": 2189
    },
    {
      "epoch": 0.6384374316740762,
      "grad_norm": 28.943283842264346,
      "learning_rate": 3.0604497286079227e-06,
      "loss": 0.7906,
      "step": 2190
    },
    {
      "epoch": 0.6387289556154799,
      "grad_norm": 23.993107489348247,
      "learning_rate": 3.0561002369782865e-06,
      "loss": 0.719,
      "step": 2191
    },
    {
      "epoch": 0.6390204795568836,
      "grad_norm": 40.43842053136658,
      "learning_rate": 3.051752477584191e-06,
      "loss": 0.8496,
      "step": 2192
    },
    {
      "epoch": 0.6393120034982873,
      "grad_norm": 40.20575276128765,
      "learning_rate": 3.0474064542999837e-06,
      "loss": 0.7002,
      "step": 2193
    },
    {
      "epoch": 0.639603527439691,
      "grad_norm": 39.61709820653485,
      "learning_rate": 3.043062170998464e-06,
      "loss": 0.6621,
      "step": 2194
    },
    {
      "epoch": 0.6398950513810947,
      "grad_norm": 26.132764918870784,
      "learning_rate": 3.0387196315508827e-06,
      "loss": 1.0757,
      "step": 2195
    },
    {
      "epoch": 0.6401865753224983,
      "grad_norm": 22.64143319319625,
      "learning_rate": 3.0343788398269342e-06,
      "loss": 0.837,
      "step": 2196
    },
    {
      "epoch": 0.6404780992639021,
      "grad_norm": 84.86063531521401,
      "learning_rate": 3.0300397996947604e-06,
      "loss": 0.7381,
      "step": 2197
    },
    {
      "epoch": 0.6407696232053057,
      "grad_norm": 22.33332546773624,
      "learning_rate": 3.025702515020937e-06,
      "loss": 0.6943,
      "step": 2198
    },
    {
      "epoch": 0.6410611471467095,
      "grad_norm": 26.485617600394722,
      "learning_rate": 3.021366989670475e-06,
      "loss": 0.8054,
      "step": 2199
    },
    {
      "epoch": 0.6413526710881131,
      "grad_norm": 37.04636303035897,
      "learning_rate": 3.0170332275068247e-06,
      "loss": 0.7978,
      "step": 2200
    },
    {
      "epoch": 0.6413526710881131,
      "eval_loss": 1.119762659072876,
      "eval_runtime": 2247.6765,
      "eval_samples_per_second": 7.683,
      "eval_steps_per_second": 0.427,
      "step": 2200
    },
    {
      "epoch": 0.6416441950295168,
      "grad_norm": 17.672349197242916,
      "learning_rate": 3.012701232391857e-06,
      "loss": 0.6857,
      "step": 2201
    },
    {
      "epoch": 0.6419357189709205,
      "grad_norm": 17.265047809663624,
      "learning_rate": 3.0083710081858748e-06,
      "loss": 0.728,
      "step": 2202
    },
    {
      "epoch": 0.6422272429123241,
      "grad_norm": 39.96052301502084,
      "learning_rate": 3.004042558747598e-06,
      "loss": 0.7073,
      "step": 2203
    },
    {
      "epoch": 0.6425187668537279,
      "grad_norm": 39.00800750083241,
      "learning_rate": 2.9997158879341647e-06,
      "loss": 0.8275,
      "step": 2204
    },
    {
      "epoch": 0.6428102907951315,
      "grad_norm": 23.261074632072358,
      "learning_rate": 2.995390999601135e-06,
      "loss": 0.732,
      "step": 2205
    },
    {
      "epoch": 0.6431018147365353,
      "grad_norm": 24.4118363262218,
      "learning_rate": 2.9910678976024733e-06,
      "loss": 0.8509,
      "step": 2206
    },
    {
      "epoch": 0.6433933386779389,
      "grad_norm": 20.316980799040042,
      "learning_rate": 2.9867465857905524e-06,
      "loss": 0.838,
      "step": 2207
    },
    {
      "epoch": 0.6436848626193427,
      "grad_norm": 108.91856107916983,
      "learning_rate": 2.982427068016155e-06,
      "loss": 0.6845,
      "step": 2208
    },
    {
      "epoch": 0.6439763865607463,
      "grad_norm": 22.50827667227923,
      "learning_rate": 2.978109348128462e-06,
      "loss": 0.8439,
      "step": 2209
    },
    {
      "epoch": 0.64426791050215,
      "grad_norm": 16.552301753331527,
      "learning_rate": 2.9737934299750514e-06,
      "loss": 0.5738,
      "step": 2210
    },
    {
      "epoch": 0.6445594344435537,
      "grad_norm": 14.969596472282628,
      "learning_rate": 2.9694793174018976e-06,
      "loss": 0.8944,
      "step": 2211
    },
    {
      "epoch": 0.6448509583849573,
      "grad_norm": 46.85984588576351,
      "learning_rate": 2.965167014253363e-06,
      "loss": 0.7511,
      "step": 2212
    },
    {
      "epoch": 0.6451424823263611,
      "grad_norm": 17.397323235173346,
      "learning_rate": 2.960856524372202e-06,
      "loss": 0.6468,
      "step": 2213
    },
    {
      "epoch": 0.6454340062677647,
      "grad_norm": 24.497095993759,
      "learning_rate": 2.956547851599548e-06,
      "loss": 0.8638,
      "step": 2214
    },
    {
      "epoch": 0.6457255302091685,
      "grad_norm": 23.565782187029292,
      "learning_rate": 2.952240999774923e-06,
      "loss": 0.7828,
      "step": 2215
    },
    {
      "epoch": 0.6460170541505721,
      "grad_norm": 18.998886974943755,
      "learning_rate": 2.947935972736217e-06,
      "loss": 0.8972,
      "step": 2216
    },
    {
      "epoch": 0.6463085780919758,
      "grad_norm": 24.305922953984822,
      "learning_rate": 2.943632774319698e-06,
      "loss": 0.7808,
      "step": 2217
    },
    {
      "epoch": 0.6466001020333795,
      "grad_norm": 22.23103378279549,
      "learning_rate": 2.9393314083600076e-06,
      "loss": 0.7471,
      "step": 2218
    },
    {
      "epoch": 0.6468916259747832,
      "grad_norm": 82.29547930644374,
      "learning_rate": 2.93503187869015e-06,
      "loss": 0.7436,
      "step": 2219
    },
    {
      "epoch": 0.6471831499161869,
      "grad_norm": 36.62922733744492,
      "learning_rate": 2.930734189141492e-06,
      "loss": 0.7956,
      "step": 2220
    },
    {
      "epoch": 0.6474746738575905,
      "grad_norm": 27.063333418991093,
      "learning_rate": 2.926438343543769e-06,
      "loss": 0.6958,
      "step": 2221
    },
    {
      "epoch": 0.6477661977989942,
      "grad_norm": 39.18125995431012,
      "learning_rate": 2.922144345725062e-06,
      "loss": 0.697,
      "step": 2222
    },
    {
      "epoch": 0.6480577217403979,
      "grad_norm": 26.748009942648675,
      "learning_rate": 2.9178521995118154e-06,
      "loss": 0.8698,
      "step": 2223
    },
    {
      "epoch": 0.6483492456818016,
      "grad_norm": 53.693362868360296,
      "learning_rate": 2.9135619087288153e-06,
      "loss": 0.6488,
      "step": 2224
    },
    {
      "epoch": 0.6486407696232053,
      "grad_norm": 36.16666004340089,
      "learning_rate": 2.9092734771991998e-06,
      "loss": 0.6416,
      "step": 2225
    },
    {
      "epoch": 0.648932293564609,
      "grad_norm": 19.57966261891017,
      "learning_rate": 2.9049869087444493e-06,
      "loss": 0.8194,
      "step": 2226
    },
    {
      "epoch": 0.6492238175060127,
      "grad_norm": 17.965622938354198,
      "learning_rate": 2.9007022071843814e-06,
      "loss": 0.6484,
      "step": 2227
    },
    {
      "epoch": 0.6495153414474164,
      "grad_norm": 34.67268386469033,
      "learning_rate": 2.8964193763371546e-06,
      "loss": 0.8164,
      "step": 2228
    },
    {
      "epoch": 0.64980686538882,
      "grad_norm": 26.664312911090175,
      "learning_rate": 2.892138420019258e-06,
      "loss": 0.8657,
      "step": 2229
    },
    {
      "epoch": 0.6500983893302238,
      "grad_norm": 32.33390644149627,
      "learning_rate": 2.887859342045506e-06,
      "loss": 0.8506,
      "step": 2230
    },
    {
      "epoch": 0.6503899132716274,
      "grad_norm": 63.780589909755335,
      "learning_rate": 2.883582146229046e-06,
      "loss": 0.8077,
      "step": 2231
    },
    {
      "epoch": 0.6506814372130311,
      "grad_norm": 38.93459321929143,
      "learning_rate": 2.879306836381345e-06,
      "loss": 0.5817,
      "step": 2232
    },
    {
      "epoch": 0.6509729611544348,
      "grad_norm": 16.923419015694765,
      "learning_rate": 2.8750334163121906e-06,
      "loss": 0.7847,
      "step": 2233
    },
    {
      "epoch": 0.6512644850958385,
      "grad_norm": 33.49926935514873,
      "learning_rate": 2.8707618898296864e-06,
      "loss": 0.6882,
      "step": 2234
    },
    {
      "epoch": 0.6515560090372422,
      "grad_norm": 37.524085507645076,
      "learning_rate": 2.8664922607402423e-06,
      "loss": 0.9373,
      "step": 2235
    },
    {
      "epoch": 0.6518475329786458,
      "grad_norm": 30.154881080177493,
      "learning_rate": 2.862224532848591e-06,
      "loss": 0.7335,
      "step": 2236
    },
    {
      "epoch": 0.6521390569200496,
      "grad_norm": 26.53340118304353,
      "learning_rate": 2.8579587099577567e-06,
      "loss": 0.8687,
      "step": 2237
    },
    {
      "epoch": 0.6524305808614532,
      "grad_norm": 16.592550360148127,
      "learning_rate": 2.853694795869074e-06,
      "loss": 0.664,
      "step": 2238
    },
    {
      "epoch": 0.652722104802857,
      "grad_norm": 25.24352695710812,
      "learning_rate": 2.8494327943821757e-06,
      "loss": 0.7029,
      "step": 2239
    },
    {
      "epoch": 0.6530136287442606,
      "grad_norm": 26.065985116543814,
      "learning_rate": 2.845172709294989e-06,
      "loss": 0.9089,
      "step": 2240
    },
    {
      "epoch": 0.6533051526856644,
      "grad_norm": 21.4280176567642,
      "learning_rate": 2.840914544403733e-06,
      "loss": 0.7768,
      "step": 2241
    },
    {
      "epoch": 0.653596676627068,
      "grad_norm": 23.788355334692916,
      "learning_rate": 2.8366583035029194e-06,
      "loss": 0.65,
      "step": 2242
    },
    {
      "epoch": 0.6538882005684716,
      "grad_norm": 97.13780090761591,
      "learning_rate": 2.8324039903853357e-06,
      "loss": 0.7649,
      "step": 2243
    },
    {
      "epoch": 0.6541797245098754,
      "grad_norm": 25.76515917968915,
      "learning_rate": 2.8281516088420665e-06,
      "loss": 0.7013,
      "step": 2244
    },
    {
      "epoch": 0.654471248451279,
      "grad_norm": 16.95339324315199,
      "learning_rate": 2.82390116266246e-06,
      "loss": 0.7448,
      "step": 2245
    },
    {
      "epoch": 0.6547627723926828,
      "grad_norm": 33.0483019732093,
      "learning_rate": 2.819652655634151e-06,
      "loss": 0.7354,
      "step": 2246
    },
    {
      "epoch": 0.6550542963340864,
      "grad_norm": 22.238264802503785,
      "learning_rate": 2.815406091543039e-06,
      "loss": 0.8336,
      "step": 2247
    },
    {
      "epoch": 0.6553458202754902,
      "grad_norm": 22.247826269375782,
      "learning_rate": 2.8111614741732975e-06,
      "loss": 0.6702,
      "step": 2248
    },
    {
      "epoch": 0.6556373442168938,
      "grad_norm": 23.807675775131433,
      "learning_rate": 2.806918807307363e-06,
      "loss": 0.7021,
      "step": 2249
    },
    {
      "epoch": 0.6559288681582975,
      "grad_norm": 25.522419014725823,
      "learning_rate": 2.802678094725931e-06,
      "loss": 0.9744,
      "step": 2250
    },
    {
      "epoch": 0.6559288681582975,
      "eval_loss": 1.1176626682281494,
      "eval_runtime": 2234.2615,
      "eval_samples_per_second": 7.73,
      "eval_steps_per_second": 0.43,
      "step": 2250
    },
    {
      "epoch": 0.6562203920997012,
      "grad_norm": 24.365616060613362,
      "learning_rate": 2.7984393402079603e-06,
      "loss": 0.7749,
      "step": 2251
    },
    {
      "epoch": 0.6565119160411049,
      "grad_norm": 41.14407075600137,
      "learning_rate": 2.794202547530661e-06,
      "loss": 0.6845,
      "step": 2252
    },
    {
      "epoch": 0.6568034399825086,
      "grad_norm": 17.239852183787406,
      "learning_rate": 2.7899677204694987e-06,
      "loss": 0.6723,
      "step": 2253
    },
    {
      "epoch": 0.6570949639239122,
      "grad_norm": 18.53330249389467,
      "learning_rate": 2.785734862798184e-06,
      "loss": 0.9217,
      "step": 2254
    },
    {
      "epoch": 0.657386487865316,
      "grad_norm": 23.045588498448893,
      "learning_rate": 2.781503978288677e-06,
      "loss": 0.9382,
      "step": 2255
    },
    {
      "epoch": 0.6576780118067196,
      "grad_norm": 18.645321163622157,
      "learning_rate": 2.77727507071117e-06,
      "loss": 0.7968,
      "step": 2256
    },
    {
      "epoch": 0.6579695357481233,
      "grad_norm": 32.207472676490816,
      "learning_rate": 2.7730481438341083e-06,
      "loss": 0.7784,
      "step": 2257
    },
    {
      "epoch": 0.658261059689527,
      "grad_norm": 34.003368916372075,
      "learning_rate": 2.768823201424158e-06,
      "loss": 0.7694,
      "step": 2258
    },
    {
      "epoch": 0.6585525836309307,
      "grad_norm": 25.48315249329304,
      "learning_rate": 2.764600247246224e-06,
      "loss": 0.767,
      "step": 2259
    },
    {
      "epoch": 0.6588441075723344,
      "grad_norm": 18.639327822284965,
      "learning_rate": 2.7603792850634402e-06,
      "loss": 0.7677,
      "step": 2260
    },
    {
      "epoch": 0.6591356315137381,
      "grad_norm": 22.277865897882847,
      "learning_rate": 2.756160318637161e-06,
      "loss": 0.7847,
      "step": 2261
    },
    {
      "epoch": 0.6594271554551417,
      "grad_norm": 35.16458524164344,
      "learning_rate": 2.7519433517269665e-06,
      "loss": 0.8069,
      "step": 2262
    },
    {
      "epoch": 0.6597186793965455,
      "grad_norm": 19.008459234184357,
      "learning_rate": 2.7477283880906546e-06,
      "loss": 0.8104,
      "step": 2263
    },
    {
      "epoch": 0.6600102033379491,
      "grad_norm": 17.303718247595626,
      "learning_rate": 2.7435154314842337e-06,
      "loss": 0.834,
      "step": 2264
    },
    {
      "epoch": 0.6603017272793528,
      "grad_norm": 17.78128678689521,
      "learning_rate": 2.7393044856619277e-06,
      "loss": 0.8195,
      "step": 2265
    },
    {
      "epoch": 0.6605932512207565,
      "grad_norm": 23.70037106052974,
      "learning_rate": 2.7350955543761682e-06,
      "loss": 0.5991,
      "step": 2266
    },
    {
      "epoch": 0.6608847751621602,
      "grad_norm": 20.67824636158525,
      "learning_rate": 2.7308886413775905e-06,
      "loss": 0.8163,
      "step": 2267
    },
    {
      "epoch": 0.6611762991035639,
      "grad_norm": 15.464596046341665,
      "learning_rate": 2.7266837504150345e-06,
      "loss": 0.731,
      "step": 2268
    },
    {
      "epoch": 0.6614678230449675,
      "grad_norm": 36.06996707135331,
      "learning_rate": 2.7224808852355313e-06,
      "loss": 0.744,
      "step": 2269
    },
    {
      "epoch": 0.6617593469863713,
      "grad_norm": 38.62205746962612,
      "learning_rate": 2.7182800495843166e-06,
      "loss": 0.7672,
      "step": 2270
    },
    {
      "epoch": 0.6620508709277749,
      "grad_norm": 34.76737980709617,
      "learning_rate": 2.7140812472048077e-06,
      "loss": 0.7089,
      "step": 2271
    },
    {
      "epoch": 0.6623423948691787,
      "grad_norm": 25.390027353669094,
      "learning_rate": 2.7098844818386164e-06,
      "loss": 0.7494,
      "step": 2272
    },
    {
      "epoch": 0.6626339188105823,
      "grad_norm": 21.538009207556307,
      "learning_rate": 2.705689757225538e-06,
      "loss": 0.5258,
      "step": 2273
    },
    {
      "epoch": 0.6629254427519861,
      "grad_norm": 26.94294843372255,
      "learning_rate": 2.7014970771035474e-06,
      "loss": 0.9156,
      "step": 2274
    },
    {
      "epoch": 0.6632169666933897,
      "grad_norm": 24.52374058743206,
      "learning_rate": 2.6973064452088e-06,
      "loss": 0.7096,
      "step": 2275
    },
    {
      "epoch": 0.6635084906347933,
      "grad_norm": 57.36092528726927,
      "learning_rate": 2.6931178652756262e-06,
      "loss": 0.7265,
      "step": 2276
    },
    {
      "epoch": 0.6638000145761971,
      "grad_norm": 27.876328448931012,
      "learning_rate": 2.6889313410365214e-06,
      "loss": 0.8551,
      "step": 2277
    },
    {
      "epoch": 0.6640915385176007,
      "grad_norm": 20.873116592422623,
      "learning_rate": 2.6847468762221616e-06,
      "loss": 0.8265,
      "step": 2278
    },
    {
      "epoch": 0.6643830624590045,
      "grad_norm": 27.189344647140306,
      "learning_rate": 2.6805644745613748e-06,
      "loss": 0.8579,
      "step": 2279
    },
    {
      "epoch": 0.6646745864004081,
      "grad_norm": 21.859574202845202,
      "learning_rate": 2.6763841397811576e-06,
      "loss": 0.695,
      "step": 2280
    },
    {
      "epoch": 0.6649661103418119,
      "grad_norm": 16.089554511305668,
      "learning_rate": 2.672205875606664e-06,
      "loss": 0.6999,
      "step": 2281
    },
    {
      "epoch": 0.6652576342832155,
      "grad_norm": 14.198146248573531,
      "learning_rate": 2.668029685761201e-06,
      "loss": 0.8137,
      "step": 2282
    },
    {
      "epoch": 0.6655491582246192,
      "grad_norm": 41.908310021398755,
      "learning_rate": 2.6638555739662314e-06,
      "loss": 0.744,
      "step": 2283
    },
    {
      "epoch": 0.6658406821660229,
      "grad_norm": 20.84773437837355,
      "learning_rate": 2.6596835439413584e-06,
      "loss": 0.7886,
      "step": 2284
    },
    {
      "epoch": 0.6661322061074266,
      "grad_norm": 17.29229864521424,
      "learning_rate": 2.655513599404337e-06,
      "loss": 0.6315,
      "step": 2285
    },
    {
      "epoch": 0.6664237300488303,
      "grad_norm": 28.703296463149176,
      "learning_rate": 2.6513457440710612e-06,
      "loss": 0.614,
      "step": 2286
    },
    {
      "epoch": 0.6667152539902339,
      "grad_norm": 21.760203636423565,
      "learning_rate": 2.6471799816555632e-06,
      "loss": 0.6835,
      "step": 2287
    },
    {
      "epoch": 0.6670067779316377,
      "grad_norm": 23.140112368997745,
      "learning_rate": 2.6430163158700116e-06,
      "loss": 0.7833,
      "step": 2288
    },
    {
      "epoch": 0.6672983018730413,
      "grad_norm": 86.11724819624257,
      "learning_rate": 2.6388547504247076e-06,
      "loss": 0.8838,
      "step": 2289
    },
    {
      "epoch": 0.667589825814445,
      "grad_norm": 16.533841978839266,
      "learning_rate": 2.634695289028072e-06,
      "loss": 0.6259,
      "step": 2290
    },
    {
      "epoch": 0.6678813497558487,
      "grad_norm": 32.32478205219583,
      "learning_rate": 2.630537935386666e-06,
      "loss": 0.7638,
      "step": 2291
    },
    {
      "epoch": 0.6681728736972524,
      "grad_norm": 23.858612110992187,
      "learning_rate": 2.6263826932051562e-06,
      "loss": 0.7161,
      "step": 2292
    },
    {
      "epoch": 0.6684643976386561,
      "grad_norm": 22.392054178239615,
      "learning_rate": 2.622229566186343e-06,
      "loss": 0.7471,
      "step": 2293
    },
    {
      "epoch": 0.6687559215800598,
      "grad_norm": 54.74650436827159,
      "learning_rate": 2.6180785580311284e-06,
      "loss": 0.7291,
      "step": 2294
    },
    {
      "epoch": 0.6690474455214634,
      "grad_norm": 47.464204156738425,
      "learning_rate": 2.6139296724385354e-06,
      "loss": 0.7401,
      "step": 2295
    },
    {
      "epoch": 0.6693389694628671,
      "grad_norm": 26.828720668927726,
      "learning_rate": 2.609782913105691e-06,
      "loss": 0.7079,
      "step": 2296
    },
    {
      "epoch": 0.6696304934042708,
      "grad_norm": 34.09890707150945,
      "learning_rate": 2.605638283727831e-06,
      "loss": 0.8854,
      "step": 2297
    },
    {
      "epoch": 0.6699220173456745,
      "grad_norm": 27.083223674755605,
      "learning_rate": 2.601495787998288e-06,
      "loss": 0.8338,
      "step": 2298
    },
    {
      "epoch": 0.6702135412870782,
      "grad_norm": 41.02656019975677,
      "learning_rate": 2.5973554296084978e-06,
      "loss": 0.8327,
      "step": 2299
    },
    {
      "epoch": 0.6705050652284819,
      "grad_norm": 16.222296151243615,
      "learning_rate": 2.59321721224799e-06,
      "loss": 0.6784,
      "step": 2300
    },
    {
      "epoch": 0.6705050652284819,
      "eval_loss": 1.116878867149353,
      "eval_runtime": 2176.2044,
      "eval_samples_per_second": 7.936,
      "eval_steps_per_second": 0.441,
      "step": 2300
    },
    {
      "epoch": 0.6707965891698856,
      "grad_norm": 23.975561708434217,
      "learning_rate": 2.589081139604387e-06,
      "loss": 0.7848,
      "step": 2301
    },
    {
      "epoch": 0.6710881131112892,
      "grad_norm": 29.152576128729887,
      "learning_rate": 2.5849472153634003e-06,
      "loss": 0.7204,
      "step": 2302
    },
    {
      "epoch": 0.671379637052693,
      "grad_norm": 36.75254865303604,
      "learning_rate": 2.5808154432088204e-06,
      "loss": 0.7658,
      "step": 2303
    },
    {
      "epoch": 0.6716711609940966,
      "grad_norm": 36.89783177448001,
      "learning_rate": 2.576685826822535e-06,
      "loss": 0.7594,
      "step": 2304
    },
    {
      "epoch": 0.6719626849355004,
      "grad_norm": 21.153339525489482,
      "learning_rate": 2.5725583698844924e-06,
      "loss": 0.7024,
      "step": 2305
    },
    {
      "epoch": 0.672254208876904,
      "grad_norm": 28.79769403409019,
      "learning_rate": 2.568433076072734e-06,
      "loss": 0.7016,
      "step": 2306
    },
    {
      "epoch": 0.6725457328183077,
      "grad_norm": 36.7857257461711,
      "learning_rate": 2.5643099490633582e-06,
      "loss": 0.7454,
      "step": 2307
    },
    {
      "epoch": 0.6728372567597114,
      "grad_norm": 27.04606411376863,
      "learning_rate": 2.5601889925305433e-06,
      "loss": 0.7455,
      "step": 2308
    },
    {
      "epoch": 0.673128780701115,
      "grad_norm": 33.63779535645963,
      "learning_rate": 2.5560702101465283e-06,
      "loss": 0.8524,
      "step": 2309
    },
    {
      "epoch": 0.6734203046425188,
      "grad_norm": 25.558414713686442,
      "learning_rate": 2.5519536055816194e-06,
      "loss": 0.7164,
      "step": 2310
    },
    {
      "epoch": 0.6737118285839224,
      "grad_norm": 24.397891317540775,
      "learning_rate": 2.547839182504175e-06,
      "loss": 0.7,
      "step": 2311
    },
    {
      "epoch": 0.6740033525253262,
      "grad_norm": 20.817704799005565,
      "learning_rate": 2.5437269445806146e-06,
      "loss": 0.6963,
      "step": 2312
    },
    {
      "epoch": 0.6742948764667298,
      "grad_norm": 116.53635520707836,
      "learning_rate": 2.5396168954754093e-06,
      "loss": 0.6644,
      "step": 2313
    },
    {
      "epoch": 0.6745864004081336,
      "grad_norm": 27.67954990619317,
      "learning_rate": 2.5355090388510806e-06,
      "loss": 0.9206,
      "step": 2314
    },
    {
      "epoch": 0.6748779243495372,
      "grad_norm": 22.008052925266647,
      "learning_rate": 2.531403378368197e-06,
      "loss": 0.6377,
      "step": 2315
    },
    {
      "epoch": 0.6751694482909409,
      "grad_norm": 15.696222763025766,
      "learning_rate": 2.527299917685362e-06,
      "loss": 0.6822,
      "step": 2316
    },
    {
      "epoch": 0.6754609722323446,
      "grad_norm": 25.69104130304068,
      "learning_rate": 2.5231986604592346e-06,
      "loss": 0.7098,
      "step": 2317
    },
    {
      "epoch": 0.6757524961737482,
      "grad_norm": 32.025438535991924,
      "learning_rate": 2.519099610344492e-06,
      "loss": 0.842,
      "step": 2318
    },
    {
      "epoch": 0.676044020115152,
      "grad_norm": 23.572145217525033,
      "learning_rate": 2.5150027709938616e-06,
      "loss": 0.7405,
      "step": 2319
    },
    {
      "epoch": 0.6763355440565556,
      "grad_norm": 41.93629752112546,
      "learning_rate": 2.5109081460580875e-06,
      "loss": 0.6404,
      "step": 2320
    },
    {
      "epoch": 0.6766270679979594,
      "grad_norm": 31.148958408689452,
      "learning_rate": 2.506815739185946e-06,
      "loss": 0.8218,
      "step": 2321
    },
    {
      "epoch": 0.676918591939363,
      "grad_norm": 23.97788159671946,
      "learning_rate": 2.502725554024239e-06,
      "loss": 0.7676,
      "step": 2322
    },
    {
      "epoch": 0.6772101158807667,
      "grad_norm": 35.545559180297566,
      "learning_rate": 2.4986375942177856e-06,
      "loss": 0.7571,
      "step": 2323
    },
    {
      "epoch": 0.6775016398221704,
      "grad_norm": 23.638537066112242,
      "learning_rate": 2.494551863409418e-06,
      "loss": 0.8071,
      "step": 2324
    },
    {
      "epoch": 0.6777931637635741,
      "grad_norm": 14.720005264431414,
      "learning_rate": 2.490468365239994e-06,
      "loss": 0.783,
      "step": 2325
    },
    {
      "epoch": 0.6780846877049778,
      "grad_norm": 86.09176089822532,
      "learning_rate": 2.4863871033483693e-06,
      "loss": 0.6981,
      "step": 2326
    },
    {
      "epoch": 0.6783762116463815,
      "grad_norm": 40.676609655869115,
      "learning_rate": 2.482308081371413e-06,
      "loss": 0.6631,
      "step": 2327
    },
    {
      "epoch": 0.6786677355877851,
      "grad_norm": 18.02141739905531,
      "learning_rate": 2.478231302943997e-06,
      "loss": 0.6857,
      "step": 2328
    },
    {
      "epoch": 0.6789592595291888,
      "grad_norm": 22.794605678008853,
      "learning_rate": 2.4741567716989944e-06,
      "loss": 0.7052,
      "step": 2329
    },
    {
      "epoch": 0.6792507834705925,
      "grad_norm": 39.04209415539729,
      "learning_rate": 2.470084491267278e-06,
      "loss": 0.7392,
      "step": 2330
    },
    {
      "epoch": 0.6795423074119962,
      "grad_norm": 32.397496880306775,
      "learning_rate": 2.4660144652777053e-06,
      "loss": 0.7477,
      "step": 2331
    },
    {
      "epoch": 0.6798338313533999,
      "grad_norm": 22.316211157606144,
      "learning_rate": 2.46194669735714e-06,
      "loss": 0.5768,
      "step": 2332
    },
    {
      "epoch": 0.6801253552948036,
      "grad_norm": 25.19484198086135,
      "learning_rate": 2.45788119113042e-06,
      "loss": 0.8632,
      "step": 2333
    },
    {
      "epoch": 0.6804168792362073,
      "grad_norm": 31.91911534776849,
      "learning_rate": 2.4538179502203753e-06,
      "loss": 0.8089,
      "step": 2334
    },
    {
      "epoch": 0.6807084031776109,
      "grad_norm": 20.134259619154065,
      "learning_rate": 2.449756978247815e-06,
      "loss": 0.8037,
      "step": 2335
    },
    {
      "epoch": 0.6809999271190147,
      "grad_norm": 19.187884252937444,
      "learning_rate": 2.445698278831528e-06,
      "loss": 0.7084,
      "step": 2336
    },
    {
      "epoch": 0.6812914510604183,
      "grad_norm": 36.691533494956374,
      "learning_rate": 2.4416418555882715e-06,
      "loss": 0.6917,
      "step": 2337
    },
    {
      "epoch": 0.6815829750018221,
      "grad_norm": 23.260240425586314,
      "learning_rate": 2.437587712132787e-06,
      "loss": 0.7011,
      "step": 2338
    },
    {
      "epoch": 0.6818744989432257,
      "grad_norm": 33.982988061564285,
      "learning_rate": 2.4335358520777687e-06,
      "loss": 0.8548,
      "step": 2339
    },
    {
      "epoch": 0.6821660228846294,
      "grad_norm": 16.429114034501406,
      "learning_rate": 2.429486279033892e-06,
      "loss": 0.7066,
      "step": 2340
    },
    {
      "epoch": 0.6824575468260331,
      "grad_norm": 15.314092152940912,
      "learning_rate": 2.425438996609782e-06,
      "loss": 0.7805,
      "step": 2341
    },
    {
      "epoch": 0.6827490707674367,
      "grad_norm": 17.970855599290047,
      "learning_rate": 2.4213940084120274e-06,
      "loss": 0.7026,
      "step": 2342
    },
    {
      "epoch": 0.6830405947088405,
      "grad_norm": 21.86765551874261,
      "learning_rate": 2.417351318045173e-06,
      "loss": 0.8078,
      "step": 2343
    },
    {
      "epoch": 0.6833321186502441,
      "grad_norm": 13.547873863332931,
      "learning_rate": 2.4133109291117156e-06,
      "loss": 0.597,
      "step": 2344
    },
    {
      "epoch": 0.6836236425916479,
      "grad_norm": 29.54625436749827,
      "learning_rate": 2.409272845212102e-06,
      "loss": 0.7218,
      "step": 2345
    },
    {
      "epoch": 0.6839151665330515,
      "grad_norm": 25.74431687831202,
      "learning_rate": 2.405237069944721e-06,
      "loss": 0.7719,
      "step": 2346
    },
    {
      "epoch": 0.6842066904744553,
      "grad_norm": 20.72939316077721,
      "learning_rate": 2.4012036069059074e-06,
      "loss": 0.6601,
      "step": 2347
    },
    {
      "epoch": 0.6844982144158589,
      "grad_norm": 22.468234206812053,
      "learning_rate": 2.397172459689936e-06,
      "loss": 0.7266,
      "step": 2348
    },
    {
      "epoch": 0.6847897383572626,
      "grad_norm": 24.192897533858016,
      "learning_rate": 2.3931436318890182e-06,
      "loss": 0.6463,
      "step": 2349
    },
    {
      "epoch": 0.6850812622986663,
      "grad_norm": 26.67689474192638,
      "learning_rate": 2.3891171270932923e-06,
      "loss": 0.6842,
      "step": 2350
    },
    {
      "epoch": 0.6850812622986663,
      "eval_loss": 1.1174310445785522,
      "eval_runtime": 2181.5743,
      "eval_samples_per_second": 7.916,
      "eval_steps_per_second": 0.44,
      "step": 2350
    },
    {
      "epoch": 0.6853727862400699,
      "grad_norm": 50.078331551025016,
      "learning_rate": 2.3850929488908386e-06,
      "loss": 0.6677,
      "step": 2351
    },
    {
      "epoch": 0.6856643101814737,
      "grad_norm": 28.251906916561808,
      "learning_rate": 2.3810711008676495e-06,
      "loss": 0.8149,
      "step": 2352
    },
    {
      "epoch": 0.6859558341228773,
      "grad_norm": 25.13464248953298,
      "learning_rate": 2.3770515866076564e-06,
      "loss": 0.7829,
      "step": 2353
    },
    {
      "epoch": 0.686247358064281,
      "grad_norm": 16.205730818379553,
      "learning_rate": 2.3730344096926974e-06,
      "loss": 0.6802,
      "step": 2354
    },
    {
      "epoch": 0.6865388820056847,
      "grad_norm": 32.98925809168679,
      "learning_rate": 2.369019573702535e-06,
      "loss": 0.8492,
      "step": 2355
    },
    {
      "epoch": 0.6868304059470884,
      "grad_norm": 18.167760267009474,
      "learning_rate": 2.3650070822148447e-06,
      "loss": 0.8118,
      "step": 2356
    },
    {
      "epoch": 0.6871219298884921,
      "grad_norm": 30.405105497159795,
      "learning_rate": 2.3609969388052124e-06,
      "loss": 0.7196,
      "step": 2357
    },
    {
      "epoch": 0.6874134538298958,
      "grad_norm": 38.636878690279055,
      "learning_rate": 2.3569891470471308e-06,
      "loss": 0.7707,
      "step": 2358
    },
    {
      "epoch": 0.6877049777712995,
      "grad_norm": 23.496378235074545,
      "learning_rate": 2.3529837105120002e-06,
      "loss": 0.8153,
      "step": 2359
    },
    {
      "epoch": 0.6879965017127031,
      "grad_norm": 27.813173333009786,
      "learning_rate": 2.3489806327691156e-06,
      "loss": 0.7993,
      "step": 2360
    },
    {
      "epoch": 0.6882880256541068,
      "grad_norm": 38.836517960766436,
      "learning_rate": 2.3449799173856756e-06,
      "loss": 0.794,
      "step": 2361
    },
    {
      "epoch": 0.6885795495955105,
      "grad_norm": 22.897871673973228,
      "learning_rate": 2.3409815679267733e-06,
      "loss": 0.6935,
      "step": 2362
    },
    {
      "epoch": 0.6888710735369142,
      "grad_norm": 23.778457977342033,
      "learning_rate": 2.336985587955391e-06,
      "loss": 0.6954,
      "step": 2363
    },
    {
      "epoch": 0.6891625974783179,
      "grad_norm": 30.618807796050422,
      "learning_rate": 2.3329919810324036e-06,
      "loss": 0.6771,
      "step": 2364
    },
    {
      "epoch": 0.6894541214197216,
      "grad_norm": 18.428373897880146,
      "learning_rate": 2.3290007507165624e-06,
      "loss": 0.7607,
      "step": 2365
    },
    {
      "epoch": 0.6897456453611253,
      "grad_norm": 32.04533891369372,
      "learning_rate": 2.325011900564515e-06,
      "loss": 0.8006,
      "step": 2366
    },
    {
      "epoch": 0.690037169302529,
      "grad_norm": 27.72491683789267,
      "learning_rate": 2.3210254341307743e-06,
      "loss": 0.675,
      "step": 2367
    },
    {
      "epoch": 0.6903286932439326,
      "grad_norm": 21.166778164721975,
      "learning_rate": 2.3170413549677367e-06,
      "loss": 0.7567,
      "step": 2368
    },
    {
      "epoch": 0.6906202171853364,
      "grad_norm": 21.674868366090205,
      "learning_rate": 2.313059666625669e-06,
      "loss": 0.6916,
      "step": 2369
    },
    {
      "epoch": 0.69091174112674,
      "grad_norm": 26.40105974554142,
      "learning_rate": 2.3090803726527083e-06,
      "loss": 0.8234,
      "step": 2370
    },
    {
      "epoch": 0.6912032650681437,
      "grad_norm": 20.1195470262313,
      "learning_rate": 2.3051034765948578e-06,
      "loss": 0.8446,
      "step": 2371
    },
    {
      "epoch": 0.6914947890095474,
      "grad_norm": 19.10206652171736,
      "learning_rate": 2.301128981995985e-06,
      "loss": 0.5784,
      "step": 2372
    },
    {
      "epoch": 0.691786312950951,
      "grad_norm": 23.50708163534855,
      "learning_rate": 2.297156892397811e-06,
      "loss": 0.6798,
      "step": 2373
    },
    {
      "epoch": 0.6920778368923548,
      "grad_norm": 35.75135949203573,
      "learning_rate": 2.293187211339926e-06,
      "loss": 0.6937,
      "step": 2374
    },
    {
      "epoch": 0.6923693608337584,
      "grad_norm": 64.09327987785781,
      "learning_rate": 2.2892199423597617e-06,
      "loss": 0.7864,
      "step": 2375
    },
    {
      "epoch": 0.6926608847751622,
      "grad_norm": 21.652387671985686,
      "learning_rate": 2.2852550889926067e-06,
      "loss": 0.7162,
      "step": 2376
    },
    {
      "epoch": 0.6929524087165658,
      "grad_norm": 26.03486232092253,
      "learning_rate": 2.281292654771597e-06,
      "loss": 0.745,
      "step": 2377
    },
    {
      "epoch": 0.6932439326579696,
      "grad_norm": 17.275439747553154,
      "learning_rate": 2.2773326432277097e-06,
      "loss": 0.7764,
      "step": 2378
    },
    {
      "epoch": 0.6935354565993732,
      "grad_norm": 38.22265050939967,
      "learning_rate": 2.273375057889769e-06,
      "loss": 0.9096,
      "step": 2379
    },
    {
      "epoch": 0.693826980540777,
      "grad_norm": 31.796812098492456,
      "learning_rate": 2.2694199022844284e-06,
      "loss": 0.6752,
      "step": 2380
    },
    {
      "epoch": 0.6941185044821806,
      "grad_norm": 26.375498103327118,
      "learning_rate": 2.265467179936182e-06,
      "loss": 0.7221,
      "step": 2381
    },
    {
      "epoch": 0.6944100284235842,
      "grad_norm": 20.968436794932263,
      "learning_rate": 2.261516894367356e-06,
      "loss": 0.7347,
      "step": 2382
    },
    {
      "epoch": 0.694701552364988,
      "grad_norm": 15.692274079822738,
      "learning_rate": 2.257569049098103e-06,
      "loss": 0.6874,
      "step": 2383
    },
    {
      "epoch": 0.6949930763063916,
      "grad_norm": 16.261210928792543,
      "learning_rate": 2.2536236476464007e-06,
      "loss": 0.6565,
      "step": 2384
    },
    {
      "epoch": 0.6952846002477954,
      "grad_norm": 53.436489269492185,
      "learning_rate": 2.2496806935280535e-06,
      "loss": 0.7828,
      "step": 2385
    },
    {
      "epoch": 0.695576124189199,
      "grad_norm": 31.521669070331914,
      "learning_rate": 2.2457401902566745e-06,
      "loss": 0.7136,
      "step": 2386
    },
    {
      "epoch": 0.6958676481306028,
      "grad_norm": 22.95559081063324,
      "learning_rate": 2.2418021413437085e-06,
      "loss": 0.8165,
      "step": 2387
    },
    {
      "epoch": 0.6961591720720064,
      "grad_norm": 36.4759947859609,
      "learning_rate": 2.2378665502983976e-06,
      "loss": 0.7165,
      "step": 2388
    },
    {
      "epoch": 0.6964506960134101,
      "grad_norm": 49.578377236164634,
      "learning_rate": 2.233933420627804e-06,
      "loss": 0.7337,
      "step": 2389
    },
    {
      "epoch": 0.6967422199548138,
      "grad_norm": 19.638183618609084,
      "learning_rate": 2.2300027558367917e-06,
      "loss": 0.8188,
      "step": 2390
    },
    {
      "epoch": 0.6970337438962175,
      "grad_norm": 96.84729258929202,
      "learning_rate": 2.22607455942803e-06,
      "loss": 0.7564,
      "step": 2391
    },
    {
      "epoch": 0.6973252678376212,
      "grad_norm": 30.02753840140963,
      "learning_rate": 2.2221488349019903e-06,
      "loss": 0.7924,
      "step": 2392
    },
    {
      "epoch": 0.6976167917790248,
      "grad_norm": 33.06467678322622,
      "learning_rate": 2.218225585756935e-06,
      "loss": 0.8512,
      "step": 2393
    },
    {
      "epoch": 0.6979083157204286,
      "grad_norm": 16.45817053900763,
      "learning_rate": 2.2143048154889272e-06,
      "loss": 0.7712,
      "step": 2394
    },
    {
      "epoch": 0.6981998396618322,
      "grad_norm": 22.868434681442135,
      "learning_rate": 2.210386527591819e-06,
      "loss": 0.7799,
      "step": 2395
    },
    {
      "epoch": 0.6984913636032359,
      "grad_norm": 34.592698482864904,
      "learning_rate": 2.2064707255572494e-06,
      "loss": 0.7796,
      "step": 2396
    },
    {
      "epoch": 0.6987828875446396,
      "grad_norm": 30.7282273291989,
      "learning_rate": 2.2025574128746424e-06,
      "loss": 0.7184,
      "step": 2397
    },
    {
      "epoch": 0.6990744114860433,
      "grad_norm": 19.501347631194427,
      "learning_rate": 2.1986465930312067e-06,
      "loss": 0.7363,
      "step": 2398
    },
    {
      "epoch": 0.699365935427447,
      "grad_norm": 25.653408941140288,
      "learning_rate": 2.194738269511922e-06,
      "loss": 0.5961,
      "step": 2399
    },
    {
      "epoch": 0.6996574593688507,
      "grad_norm": 38.938273393939745,
      "learning_rate": 2.1908324457995556e-06,
      "loss": 0.908,
      "step": 2400
    },
    {
      "epoch": 0.6996574593688507,
      "eval_loss": 1.1167093515396118,
      "eval_runtime": 2177.1254,
      "eval_samples_per_second": 7.932,
      "eval_steps_per_second": 0.441,
      "step": 2400
    },
    {
      "epoch": 0.6999489833102543,
      "grad_norm": 26.63334491392453,
      "learning_rate": 2.186929125374634e-06,
      "loss": 0.7531,
      "step": 2401
    },
    {
      "epoch": 0.7002405072516581,
      "grad_norm": 14.054903931299647,
      "learning_rate": 2.1830283117154616e-06,
      "loss": 0.7247,
      "step": 2402
    },
    {
      "epoch": 0.7005320311930617,
      "grad_norm": 95.15099865698063,
      "learning_rate": 2.1791300082981065e-06,
      "loss": 0.6646,
      "step": 2403
    },
    {
      "epoch": 0.7008235551344654,
      "grad_norm": 28.08123603743363,
      "learning_rate": 2.1752342185964003e-06,
      "loss": 0.8089,
      "step": 2404
    },
    {
      "epoch": 0.7011150790758691,
      "grad_norm": 21.598246321722208,
      "learning_rate": 2.171340946081934e-06,
      "loss": 0.844,
      "step": 2405
    },
    {
      "epoch": 0.7014066030172728,
      "grad_norm": 20.950410084497634,
      "learning_rate": 2.1674501942240567e-06,
      "loss": 0.617,
      "step": 2406
    },
    {
      "epoch": 0.7016981269586765,
      "grad_norm": 24.425137233337725,
      "learning_rate": 2.163561966489868e-06,
      "loss": 0.823,
      "step": 2407
    },
    {
      "epoch": 0.7019896509000801,
      "grad_norm": 20.217640148045415,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.7028,
      "step": 2408
    },
    {
      "epoch": 0.7022811748414839,
      "grad_norm": 20.33881428589582,
      "learning_rate": 2.155793097249718e-06,
      "loss": 0.8274,
      "step": 2409
    },
    {
      "epoch": 0.7025726987828875,
      "grad_norm": 16.317667007446108,
      "learning_rate": 2.151912462666703e-06,
      "loss": 0.7065,
      "step": 2410
    },
    {
      "epoch": 0.7028642227242913,
      "grad_norm": 27.904529970658924,
      "learning_rate": 2.1480343660532622e-06,
      "loss": 0.7568,
      "step": 2411
    },
    {
      "epoch": 0.7031557466656949,
      "grad_norm": 20.518665663379423,
      "learning_rate": 2.144158810865217e-06,
      "loss": 0.709,
      "step": 2412
    },
    {
      "epoch": 0.7034472706070987,
      "grad_norm": 20.044116005385245,
      "learning_rate": 2.1402858005561345e-06,
      "loss": 0.6993,
      "step": 2413
    },
    {
      "epoch": 0.7037387945485023,
      "grad_norm": 26.920011907523868,
      "learning_rate": 2.1364153385773007e-06,
      "loss": 0.7976,
      "step": 2414
    },
    {
      "epoch": 0.7040303184899059,
      "grad_norm": 77.08007332974476,
      "learning_rate": 2.132547428377739e-06,
      "loss": 0.6017,
      "step": 2415
    },
    {
      "epoch": 0.7043218424313097,
      "grad_norm": 83.98671153046459,
      "learning_rate": 2.128682073404197e-06,
      "loss": 0.7524,
      "step": 2416
    },
    {
      "epoch": 0.7046133663727133,
      "grad_norm": 14.991735690311968,
      "learning_rate": 2.124819277101145e-06,
      "loss": 0.7084,
      "step": 2417
    },
    {
      "epoch": 0.7049048903141171,
      "grad_norm": 29.723050778903357,
      "learning_rate": 2.1209590429107734e-06,
      "loss": 0.7601,
      "step": 2418
    },
    {
      "epoch": 0.7051964142555207,
      "grad_norm": 20.76076785054477,
      "learning_rate": 2.1171013742729912e-06,
      "loss": 0.6491,
      "step": 2419
    },
    {
      "epoch": 0.7054879381969245,
      "grad_norm": 31.232166384761882,
      "learning_rate": 2.1132462746254147e-06,
      "loss": 0.6305,
      "step": 2420
    },
    {
      "epoch": 0.7057794621383281,
      "grad_norm": 17.357111326158073,
      "learning_rate": 2.1093937474033816e-06,
      "loss": 0.7912,
      "step": 2421
    },
    {
      "epoch": 0.7060709860797318,
      "grad_norm": 26.58736028225184,
      "learning_rate": 2.1055437960399266e-06,
      "loss": 0.7521,
      "step": 2422
    },
    {
      "epoch": 0.7063625100211355,
      "grad_norm": 14.895590691129732,
      "learning_rate": 2.1016964239657954e-06,
      "loss": 0.6564,
      "step": 2423
    },
    {
      "epoch": 0.7066540339625391,
      "grad_norm": 22.933990170127334,
      "learning_rate": 2.0978516346094342e-06,
      "loss": 0.7393,
      "step": 2424
    },
    {
      "epoch": 0.7069455579039429,
      "grad_norm": 25.923158325790467,
      "learning_rate": 2.0940094313969868e-06,
      "loss": 0.8031,
      "step": 2425
    },
    {
      "epoch": 0.7072370818453465,
      "grad_norm": 19.693110038654314,
      "learning_rate": 2.0901698177522944e-06,
      "loss": 0.8969,
      "step": 2426
    },
    {
      "epoch": 0.7075286057867503,
      "grad_norm": 25.09057746230872,
      "learning_rate": 2.0863327970968854e-06,
      "loss": 0.7568,
      "step": 2427
    },
    {
      "epoch": 0.7078201297281539,
      "grad_norm": 39.87305393321838,
      "learning_rate": 2.082498372849983e-06,
      "loss": 0.8378,
      "step": 2428
    },
    {
      "epoch": 0.7081116536695576,
      "grad_norm": 33.43735321828959,
      "learning_rate": 2.0786665484284952e-06,
      "loss": 0.6964,
      "step": 2429
    },
    {
      "epoch": 0.7084031776109613,
      "grad_norm": 27.32845157225916,
      "learning_rate": 2.074837327247012e-06,
      "loss": 0.8036,
      "step": 2430
    },
    {
      "epoch": 0.708694701552365,
      "grad_norm": 21.639783547077535,
      "learning_rate": 2.071010712717806e-06,
      "loss": 0.728,
      "step": 2431
    },
    {
      "epoch": 0.7089862254937687,
      "grad_norm": 28.314236656150342,
      "learning_rate": 2.067186708250826e-06,
      "loss": 0.683,
      "step": 2432
    },
    {
      "epoch": 0.7092777494351724,
      "grad_norm": 38.8360540855587,
      "learning_rate": 2.063365317253689e-06,
      "loss": 0.6804,
      "step": 2433
    },
    {
      "epoch": 0.709569273376576,
      "grad_norm": 23.48533497130028,
      "learning_rate": 2.059546543131696e-06,
      "loss": 0.7128,
      "step": 2434
    },
    {
      "epoch": 0.7098607973179797,
      "grad_norm": 17.06560357833926,
      "learning_rate": 2.055730389287803e-06,
      "loss": 0.7744,
      "step": 2435
    },
    {
      "epoch": 0.7101523212593834,
      "grad_norm": 19.901287985406736,
      "learning_rate": 2.051916859122641e-06,
      "loss": 0.891,
      "step": 2436
    },
    {
      "epoch": 0.7104438452007871,
      "grad_norm": 25.017037650831284,
      "learning_rate": 2.0481059560344953e-06,
      "loss": 0.6404,
      "step": 2437
    },
    {
      "epoch": 0.7107353691421908,
      "grad_norm": 24.214755500329204,
      "learning_rate": 2.0442976834193146e-06,
      "loss": 0.7219,
      "step": 2438
    },
    {
      "epoch": 0.7110268930835945,
      "grad_norm": 14.216231106869262,
      "learning_rate": 2.040492044670702e-06,
      "loss": 0.6314,
      "step": 2439
    },
    {
      "epoch": 0.7113184170249982,
      "grad_norm": 18.086976733843485,
      "learning_rate": 2.036689043179917e-06,
      "loss": 0.8469,
      "step": 2440
    },
    {
      "epoch": 0.7116099409664018,
      "grad_norm": 41.40056019009419,
      "learning_rate": 2.032888682335862e-06,
      "loss": 0.794,
      "step": 2441
    },
    {
      "epoch": 0.7119014649078056,
      "grad_norm": 36.55197055143028,
      "learning_rate": 2.0290909655250913e-06,
      "loss": 0.6668,
      "step": 2442
    },
    {
      "epoch": 0.7121929888492092,
      "grad_norm": 25.234251843083705,
      "learning_rate": 2.0252958961318024e-06,
      "loss": 0.7827,
      "step": 2443
    },
    {
      "epoch": 0.712484512790613,
      "grad_norm": 19.92605705113047,
      "learning_rate": 2.0215034775378336e-06,
      "loss": 0.8457,
      "step": 2444
    },
    {
      "epoch": 0.7127760367320166,
      "grad_norm": 22.626845779879766,
      "learning_rate": 2.0177137131226625e-06,
      "loss": 0.6797,
      "step": 2445
    },
    {
      "epoch": 0.7130675606734203,
      "grad_norm": 21.0816860380921,
      "learning_rate": 2.013926606263394e-06,
      "loss": 0.7835,
      "step": 2446
    },
    {
      "epoch": 0.713359084614824,
      "grad_norm": 40.1912148384974,
      "learning_rate": 2.010142160334778e-06,
      "loss": 0.7194,
      "step": 2447
    },
    {
      "epoch": 0.7136506085562276,
      "grad_norm": 25.771505117442096,
      "learning_rate": 2.0063603787091788e-06,
      "loss": 0.7377,
      "step": 2448
    },
    {
      "epoch": 0.7139421324976314,
      "grad_norm": 25.578097471066553,
      "learning_rate": 2.0025812647566006e-06,
      "loss": 0.6408,
      "step": 2449
    },
    {
      "epoch": 0.714233656439035,
      "grad_norm": 40.57879928148379,
      "learning_rate": 1.9988048218446577e-06,
      "loss": 0.571,
      "step": 2450
    },
    {
      "epoch": 0.714233656439035,
      "eval_loss": 1.1127010583877563,
      "eval_runtime": 2180.9119,
      "eval_samples_per_second": 7.919,
      "eval_steps_per_second": 0.44,
      "step": 2450
    },
    {
      "epoch": 0.7145251803804388,
      "grad_norm": 37.37744923924511,
      "learning_rate": 1.995031053338592e-06,
      "loss": 0.7246,
      "step": 2451
    },
    {
      "epoch": 0.7148167043218424,
      "grad_norm": 33.53121125565911,
      "learning_rate": 1.9912599626012593e-06,
      "loss": 0.7482,
      "step": 2452
    },
    {
      "epoch": 0.7151082282632462,
      "grad_norm": 24.71998355527269,
      "learning_rate": 1.9874915529931316e-06,
      "loss": 0.8456,
      "step": 2453
    },
    {
      "epoch": 0.7153997522046498,
      "grad_norm": 59.745722480948345,
      "learning_rate": 1.9837258278722855e-06,
      "loss": 0.7229,
      "step": 2454
    },
    {
      "epoch": 0.7156912761460535,
      "grad_norm": 16.951548804733505,
      "learning_rate": 1.9799627905944164e-06,
      "loss": 0.5625,
      "step": 2455
    },
    {
      "epoch": 0.7159828000874572,
      "grad_norm": 32.241801951088554,
      "learning_rate": 1.976202444512813e-06,
      "loss": 0.612,
      "step": 2456
    },
    {
      "epoch": 0.7162743240288608,
      "grad_norm": 46.01528170101002,
      "learning_rate": 1.972444792978373e-06,
      "loss": 0.7107,
      "step": 2457
    },
    {
      "epoch": 0.7165658479702646,
      "grad_norm": 25.358467254169195,
      "learning_rate": 1.96868983933959e-06,
      "loss": 0.7627,
      "step": 2458
    },
    {
      "epoch": 0.7168573719116682,
      "grad_norm": 18.65680348978791,
      "learning_rate": 1.9649375869425548e-06,
      "loss": 0.8726,
      "step": 2459
    },
    {
      "epoch": 0.717148895853072,
      "grad_norm": 25.80660059147405,
      "learning_rate": 1.9611880391309524e-06,
      "loss": 0.6451,
      "step": 2460
    },
    {
      "epoch": 0.7174404197944756,
      "grad_norm": 44.92013415933664,
      "learning_rate": 1.95744119924605e-06,
      "loss": 0.6612,
      "step": 2461
    },
    {
      "epoch": 0.7177319437358793,
      "grad_norm": 52.680969804395275,
      "learning_rate": 1.9536970706267156e-06,
      "loss": 0.6658,
      "step": 2462
    },
    {
      "epoch": 0.718023467677283,
      "grad_norm": 27.225270798319194,
      "learning_rate": 1.949955656609387e-06,
      "loss": 0.6803,
      "step": 2463
    },
    {
      "epoch": 0.7183149916186867,
      "grad_norm": 42.61096890142364,
      "learning_rate": 1.946216960528092e-06,
      "loss": 0.7939,
      "step": 2464
    },
    {
      "epoch": 0.7186065155600904,
      "grad_norm": 23.126026255891052,
      "learning_rate": 1.942480985714432e-06,
      "loss": 0.8386,
      "step": 2465
    },
    {
      "epoch": 0.7188980395014941,
      "grad_norm": 57.43926049213268,
      "learning_rate": 1.9387477354975885e-06,
      "loss": 0.7856,
      "step": 2466
    },
    {
      "epoch": 0.7191895634428978,
      "grad_norm": 19.674870769084578,
      "learning_rate": 1.9350172132043054e-06,
      "loss": 0.7141,
      "step": 2467
    },
    {
      "epoch": 0.7194810873843014,
      "grad_norm": 31.90782975288939,
      "learning_rate": 1.9312894221589085e-06,
      "loss": 0.8721,
      "step": 2468
    },
    {
      "epoch": 0.7197726113257051,
      "grad_norm": 25.815556582628222,
      "learning_rate": 1.927564365683277e-06,
      "loss": 0.509,
      "step": 2469
    },
    {
      "epoch": 0.7200641352671088,
      "grad_norm": 59.5525183512412,
      "learning_rate": 1.9238420470968665e-06,
      "loss": 0.6692,
      "step": 2470
    },
    {
      "epoch": 0.7203556592085125,
      "grad_norm": 36.49491898510384,
      "learning_rate": 1.92012246971668e-06,
      "loss": 0.5987,
      "step": 2471
    },
    {
      "epoch": 0.7206471831499162,
      "grad_norm": 18.685945989729927,
      "learning_rate": 1.9164056368572847e-06,
      "loss": 0.693,
      "step": 2472
    },
    {
      "epoch": 0.7209387070913199,
      "grad_norm": 26.559101849957283,
      "learning_rate": 1.912691551830802e-06,
      "loss": 0.7334,
      "step": 2473
    },
    {
      "epoch": 0.7212302310327235,
      "grad_norm": 27.13472925747267,
      "learning_rate": 1.9089802179469036e-06,
      "loss": 0.7399,
      "step": 2474
    },
    {
      "epoch": 0.7215217549741273,
      "grad_norm": 20.181135623386705,
      "learning_rate": 1.9052716385128105e-06,
      "loss": 0.8391,
      "step": 2475
    },
    {
      "epoch": 0.7218132789155309,
      "grad_norm": 35.64369500946453,
      "learning_rate": 1.9015658168332863e-06,
      "loss": 0.7145,
      "step": 2476
    },
    {
      "epoch": 0.7221048028569347,
      "grad_norm": 28.857025336519797,
      "learning_rate": 1.8978627562106394e-06,
      "loss": 0.6763,
      "step": 2477
    },
    {
      "epoch": 0.7223963267983383,
      "grad_norm": 27.782180535969708,
      "learning_rate": 1.8941624599447178e-06,
      "loss": 0.6488,
      "step": 2478
    },
    {
      "epoch": 0.722687850739742,
      "grad_norm": 40.61204481991765,
      "learning_rate": 1.8904649313329088e-06,
      "loss": 0.7495,
      "step": 2479
    },
    {
      "epoch": 0.7229793746811457,
      "grad_norm": 36.36521947603577,
      "learning_rate": 1.8867701736701238e-06,
      "loss": 0.5981,
      "step": 2480
    },
    {
      "epoch": 0.7232708986225493,
      "grad_norm": 175.0985302545244,
      "learning_rate": 1.8830781902488194e-06,
      "loss": 0.6811,
      "step": 2481
    },
    {
      "epoch": 0.7235624225639531,
      "grad_norm": 26.879287569947884,
      "learning_rate": 1.8793889843589647e-06,
      "loss": 0.8653,
      "step": 2482
    },
    {
      "epoch": 0.7238539465053567,
      "grad_norm": 21.16333398617138,
      "learning_rate": 1.875702559288069e-06,
      "loss": 0.7539,
      "step": 2483
    },
    {
      "epoch": 0.7241454704467605,
      "grad_norm": 18.33754850168423,
      "learning_rate": 1.87201891832115e-06,
      "loss": 0.7409,
      "step": 2484
    },
    {
      "epoch": 0.7244369943881641,
      "grad_norm": 18.79314164201513,
      "learning_rate": 1.868338064740753e-06,
      "loss": 0.779,
      "step": 2485
    },
    {
      "epoch": 0.7247285183295679,
      "grad_norm": 24.47066393365301,
      "learning_rate": 1.8646600018269356e-06,
      "loss": 0.6528,
      "step": 2486
    },
    {
      "epoch": 0.7250200422709715,
      "grad_norm": 23.36784190114525,
      "learning_rate": 1.86098473285727e-06,
      "loss": 0.7022,
      "step": 2487
    },
    {
      "epoch": 0.7253115662123751,
      "grad_norm": 29.684756215713175,
      "learning_rate": 1.8573122611068406e-06,
      "loss": 0.8095,
      "step": 2488
    },
    {
      "epoch": 0.7256030901537789,
      "grad_norm": 162.81109169889328,
      "learning_rate": 1.8536425898482336e-06,
      "loss": 0.7166,
      "step": 2489
    },
    {
      "epoch": 0.7258946140951825,
      "grad_norm": 27.962219331543867,
      "learning_rate": 1.8499757223515442e-06,
      "loss": 0.7451,
      "step": 2490
    },
    {
      "epoch": 0.7261861380365863,
      "grad_norm": 19.12273678056773,
      "learning_rate": 1.8463116618843679e-06,
      "loss": 0.8782,
      "step": 2491
    },
    {
      "epoch": 0.7264776619779899,
      "grad_norm": 46.14391092382484,
      "learning_rate": 1.8426504117118011e-06,
      "loss": 0.8424,
      "step": 2492
    },
    {
      "epoch": 0.7267691859193937,
      "grad_norm": 27.276376561788837,
      "learning_rate": 1.838991975096428e-06,
      "loss": 0.6703,
      "step": 2493
    },
    {
      "epoch": 0.7270607098607973,
      "grad_norm": 29.094714606367706,
      "learning_rate": 1.8353363552983382e-06,
      "loss": 0.6632,
      "step": 2494
    },
    {
      "epoch": 0.727352233802201,
      "grad_norm": 19.780060983666985,
      "learning_rate": 1.8316835555750985e-06,
      "loss": 0.7218,
      "step": 2495
    },
    {
      "epoch": 0.7276437577436047,
      "grad_norm": 41.66814492418623,
      "learning_rate": 1.8280335791817733e-06,
      "loss": 0.6563,
      "step": 2496
    },
    {
      "epoch": 0.7279352816850084,
      "grad_norm": 23.94326402031656,
      "learning_rate": 1.8243864293709029e-06,
      "loss": 0.7285,
      "step": 2497
    },
    {
      "epoch": 0.7282268056264121,
      "grad_norm": 18.110138999678366,
      "learning_rate": 1.8207421093925127e-06,
      "loss": 0.7987,
      "step": 2498
    },
    {
      "epoch": 0.7285183295678157,
      "grad_norm": 22.859262931017007,
      "learning_rate": 1.8171006224941068e-06,
      "loss": 0.7914,
      "step": 2499
    },
    {
      "epoch": 0.7288098535092195,
      "grad_norm": 28.67476918143253,
      "learning_rate": 1.8134619719206624e-06,
      "loss": 0.8283,
      "step": 2500
    },
    {
      "epoch": 0.7288098535092195,
      "eval_loss": 1.1174217462539673,
      "eval_runtime": 2164.1453,
      "eval_samples_per_second": 7.98,
      "eval_steps_per_second": 0.444,
      "step": 2500
    },
    {
      "epoch": 0.7291013774506231,
      "grad_norm": 60.63814611114152,
      "learning_rate": 1.8098261609146322e-06,
      "loss": 0.8204,
      "step": 2501
    },
    {
      "epoch": 0.7293929013920268,
      "grad_norm": 24.759931355335322,
      "learning_rate": 1.8061931927159377e-06,
      "loss": 0.7528,
      "step": 2502
    },
    {
      "epoch": 0.7296844253334305,
      "grad_norm": 38.536689979965395,
      "learning_rate": 1.802563070561963e-06,
      "loss": 0.763,
      "step": 2503
    },
    {
      "epoch": 0.7299759492748342,
      "grad_norm": 37.57629105592115,
      "learning_rate": 1.7989357976875603e-06,
      "loss": 0.7927,
      "step": 2504
    },
    {
      "epoch": 0.7302674732162379,
      "grad_norm": 34.742991178127696,
      "learning_rate": 1.7953113773250437e-06,
      "loss": 0.7651,
      "step": 2505
    },
    {
      "epoch": 0.7305589971576416,
      "grad_norm": 25.114303980705856,
      "learning_rate": 1.7916898127041815e-06,
      "loss": 0.7792,
      "step": 2506
    },
    {
      "epoch": 0.7308505210990452,
      "grad_norm": 26.533528996935452,
      "learning_rate": 1.7880711070522022e-06,
      "loss": 0.6598,
      "step": 2507
    },
    {
      "epoch": 0.731142045040449,
      "grad_norm": 23.40519974830833,
      "learning_rate": 1.7844552635937784e-06,
      "loss": 0.753,
      "step": 2508
    },
    {
      "epoch": 0.7314335689818526,
      "grad_norm": 73.11039880003192,
      "learning_rate": 1.7808422855510438e-06,
      "loss": 0.7151,
      "step": 2509
    },
    {
      "epoch": 0.7317250929232563,
      "grad_norm": 51.786241245550634,
      "learning_rate": 1.7772321761435674e-06,
      "loss": 0.6065,
      "step": 2510
    },
    {
      "epoch": 0.73201661686466,
      "grad_norm": 20.4599819976267,
      "learning_rate": 1.7736249385883687e-06,
      "loss": 0.629,
      "step": 2511
    },
    {
      "epoch": 0.7323081408060637,
      "grad_norm": 26.623662427685343,
      "learning_rate": 1.7700205760999061e-06,
      "loss": 0.5233,
      "step": 2512
    },
    {
      "epoch": 0.7325996647474674,
      "grad_norm": 27.68767431709484,
      "learning_rate": 1.7664190918900758e-06,
      "loss": 0.8183,
      "step": 2513
    },
    {
      "epoch": 0.732891188688871,
      "grad_norm": 23.585105011013063,
      "learning_rate": 1.76282048916821e-06,
      "loss": 0.8012,
      "step": 2514
    },
    {
      "epoch": 0.7331827126302748,
      "grad_norm": 47.18092384883461,
      "learning_rate": 1.7592247711410738e-06,
      "loss": 0.6054,
      "step": 2515
    },
    {
      "epoch": 0.7334742365716784,
      "grad_norm": 19.493462640071815,
      "learning_rate": 1.7556319410128557e-06,
      "loss": 0.6884,
      "step": 2516
    },
    {
      "epoch": 0.7337657605130822,
      "grad_norm": 22.489656205980058,
      "learning_rate": 1.752042001985182e-06,
      "loss": 0.5784,
      "step": 2517
    },
    {
      "epoch": 0.7340572844544858,
      "grad_norm": 49.68755999028854,
      "learning_rate": 1.7484549572570913e-06,
      "loss": 0.5562,
      "step": 2518
    },
    {
      "epoch": 0.7343488083958896,
      "grad_norm": 29.127219569950626,
      "learning_rate": 1.7448708100250495e-06,
      "loss": 0.5669,
      "step": 2519
    },
    {
      "epoch": 0.7346403323372932,
      "grad_norm": 23.456863198017235,
      "learning_rate": 1.7412895634829391e-06,
      "loss": 0.6531,
      "step": 2520
    },
    {
      "epoch": 0.7349318562786968,
      "grad_norm": 45.200932587360825,
      "learning_rate": 1.7377112208220581e-06,
      "loss": 0.8593,
      "step": 2521
    },
    {
      "epoch": 0.7352233802201006,
      "grad_norm": 22.07942670439618,
      "learning_rate": 1.7341357852311175e-06,
      "loss": 0.8062,
      "step": 2522
    },
    {
      "epoch": 0.7355149041615042,
      "grad_norm": 25.731671195426596,
      "learning_rate": 1.7305632598962329e-06,
      "loss": 0.8056,
      "step": 2523
    },
    {
      "epoch": 0.735806428102908,
      "grad_norm": 19.899397900578805,
      "learning_rate": 1.726993648000933e-06,
      "loss": 0.7576,
      "step": 2524
    },
    {
      "epoch": 0.7360979520443116,
      "grad_norm": 22.078815367919123,
      "learning_rate": 1.7234269527261472e-06,
      "loss": 0.6805,
      "step": 2525
    },
    {
      "epoch": 0.7363894759857154,
      "grad_norm": 30.000233134087964,
      "learning_rate": 1.7198631772502057e-06,
      "loss": 0.7999,
      "step": 2526
    },
    {
      "epoch": 0.736680999927119,
      "grad_norm": 20.74737961079211,
      "learning_rate": 1.716302324748837e-06,
      "loss": 0.64,
      "step": 2527
    },
    {
      "epoch": 0.7369725238685227,
      "grad_norm": 26.813107450924505,
      "learning_rate": 1.7127443983951687e-06,
      "loss": 0.7484,
      "step": 2528
    },
    {
      "epoch": 0.7372640478099264,
      "grad_norm": 22.282529181624085,
      "learning_rate": 1.7091894013597104e-06,
      "loss": 0.6838,
      "step": 2529
    },
    {
      "epoch": 0.7375555717513301,
      "grad_norm": 15.79746937166548,
      "learning_rate": 1.7056373368103756e-06,
      "loss": 0.7303,
      "step": 2530
    },
    {
      "epoch": 0.7378470956927338,
      "grad_norm": 33.271477153755896,
      "learning_rate": 1.7020882079124534e-06,
      "loss": 0.7591,
      "step": 2531
    },
    {
      "epoch": 0.7381386196341374,
      "grad_norm": 31.466934613815663,
      "learning_rate": 1.6985420178286216e-06,
      "loss": 0.5826,
      "step": 2532
    },
    {
      "epoch": 0.7384301435755412,
      "grad_norm": 15.642062057236007,
      "learning_rate": 1.6949987697189384e-06,
      "loss": 0.732,
      "step": 2533
    },
    {
      "epoch": 0.7387216675169448,
      "grad_norm": 23.914807533587407,
      "learning_rate": 1.6914584667408408e-06,
      "loss": 0.71,
      "step": 2534
    },
    {
      "epoch": 0.7390131914583485,
      "grad_norm": 23.804692736196646,
      "learning_rate": 1.6879211120491407e-06,
      "loss": 0.7932,
      "step": 2535
    },
    {
      "epoch": 0.7393047153997522,
      "grad_norm": 69.79331551113752,
      "learning_rate": 1.6843867087960252e-06,
      "loss": 0.6917,
      "step": 2536
    },
    {
      "epoch": 0.7395962393411559,
      "grad_norm": 29.411426745566963,
      "learning_rate": 1.680855260131045e-06,
      "loss": 0.6469,
      "step": 2537
    },
    {
      "epoch": 0.7398877632825596,
      "grad_norm": 28.261996957372357,
      "learning_rate": 1.6773267692011242e-06,
      "loss": 0.6727,
      "step": 2538
    },
    {
      "epoch": 0.7401792872239633,
      "grad_norm": 38.67312605222223,
      "learning_rate": 1.6738012391505492e-06,
      "loss": 0.6224,
      "step": 2539
    },
    {
      "epoch": 0.740470811165367,
      "grad_norm": 35.651922965163344,
      "learning_rate": 1.6702786731209681e-06,
      "loss": 0.6637,
      "step": 2540
    },
    {
      "epoch": 0.7407623351067707,
      "grad_norm": 34.27887358827945,
      "learning_rate": 1.6667590742513878e-06,
      "loss": 0.8863,
      "step": 2541
    },
    {
      "epoch": 0.7410538590481743,
      "grad_norm": 25.06141061052845,
      "learning_rate": 1.6632424456781675e-06,
      "loss": 0.6376,
      "step": 2542
    },
    {
      "epoch": 0.741345382989578,
      "grad_norm": 22.04832997149086,
      "learning_rate": 1.659728790535028e-06,
      "loss": 0.7463,
      "step": 2543
    },
    {
      "epoch": 0.7416369069309817,
      "grad_norm": 25.249838536298025,
      "learning_rate": 1.6562181119530314e-06,
      "loss": 0.856,
      "step": 2544
    },
    {
      "epoch": 0.7419284308723854,
      "grad_norm": 22.55007050404597,
      "learning_rate": 1.652710413060592e-06,
      "loss": 0.7963,
      "step": 2545
    },
    {
      "epoch": 0.7422199548137891,
      "grad_norm": 25.13290241226976,
      "learning_rate": 1.649205696983468e-06,
      "loss": 0.7785,
      "step": 2546
    },
    {
      "epoch": 0.7425114787551927,
      "grad_norm": 23.808306644111152,
      "learning_rate": 1.6457039668447605e-06,
      "loss": 0.7586,
      "step": 2547
    },
    {
      "epoch": 0.7428030026965965,
      "grad_norm": 17.1318887626226,
      "learning_rate": 1.642205225764908e-06,
      "loss": 0.6205,
      "step": 2548
    },
    {
      "epoch": 0.7430945266380001,
      "grad_norm": 19.795177391794272,
      "learning_rate": 1.6387094768616878e-06,
      "loss": 0.9847,
      "step": 2549
    },
    {
      "epoch": 0.7433860505794039,
      "grad_norm": 28.090075519321015,
      "learning_rate": 1.635216723250206e-06,
      "loss": 0.7165,
      "step": 2550
    },
    {
      "epoch": 0.7433860505794039,
      "eval_loss": 1.116228461265564,
      "eval_runtime": 2213.088,
      "eval_samples_per_second": 7.804,
      "eval_steps_per_second": 0.434,
      "step": 2550
    },
    {
      "epoch": 0.7436775745208075,
      "grad_norm": 19.981317218045877,
      "learning_rate": 1.6317269680429077e-06,
      "loss": 0.7271,
      "step": 2551
    },
    {
      "epoch": 0.7439690984622112,
      "grad_norm": 119.47019889823629,
      "learning_rate": 1.6282402143495568e-06,
      "loss": 0.6538,
      "step": 2552
    },
    {
      "epoch": 0.7442606224036149,
      "grad_norm": 26.68064704129366,
      "learning_rate": 1.624756465277249e-06,
      "loss": 0.651,
      "step": 2553
    },
    {
      "epoch": 0.7445521463450185,
      "grad_norm": 25.52363282978133,
      "learning_rate": 1.6212757239304e-06,
      "loss": 0.6818,
      "step": 2554
    },
    {
      "epoch": 0.7448436702864223,
      "grad_norm": 21.260179382761503,
      "learning_rate": 1.6177979934107462e-06,
      "loss": 0.8027,
      "step": 2555
    },
    {
      "epoch": 0.7451351942278259,
      "grad_norm": 47.65191942045954,
      "learning_rate": 1.6143232768173428e-06,
      "loss": 0.7391,
      "step": 2556
    },
    {
      "epoch": 0.7454267181692297,
      "grad_norm": 38.56644517115848,
      "learning_rate": 1.6108515772465521e-06,
      "loss": 0.8369,
      "step": 2557
    },
    {
      "epoch": 0.7457182421106333,
      "grad_norm": 20.569895753935796,
      "learning_rate": 1.6073828977920564e-06,
      "loss": 0.6903,
      "step": 2558
    },
    {
      "epoch": 0.7460097660520371,
      "grad_norm": 42.29810453447617,
      "learning_rate": 1.6039172415448416e-06,
      "loss": 0.7298,
      "step": 2559
    },
    {
      "epoch": 0.7463012899934407,
      "grad_norm": 22.868554501818732,
      "learning_rate": 1.6004546115932023e-06,
      "loss": 0.7521,
      "step": 2560
    },
    {
      "epoch": 0.7465928139348444,
      "grad_norm": 26.316314500541157,
      "learning_rate": 1.596995011022735e-06,
      "loss": 0.8251,
      "step": 2561
    },
    {
      "epoch": 0.7468843378762481,
      "grad_norm": 30.437967173039855,
      "learning_rate": 1.5935384429163376e-06,
      "loss": 0.6426,
      "step": 2562
    },
    {
      "epoch": 0.7471758618176517,
      "grad_norm": 24.336257379321058,
      "learning_rate": 1.590084910354201e-06,
      "loss": 0.7616,
      "step": 2563
    },
    {
      "epoch": 0.7474673857590555,
      "grad_norm": 92.82247147785611,
      "learning_rate": 1.5866344164138214e-06,
      "loss": 0.739,
      "step": 2564
    },
    {
      "epoch": 0.7477589097004591,
      "grad_norm": 42.41450106380287,
      "learning_rate": 1.5831869641699759e-06,
      "loss": 0.6858,
      "step": 2565
    },
    {
      "epoch": 0.7480504336418629,
      "grad_norm": 48.30339493831172,
      "learning_rate": 1.5797425566947378e-06,
      "loss": 0.8571,
      "step": 2566
    },
    {
      "epoch": 0.7483419575832665,
      "grad_norm": 25.067769143840717,
      "learning_rate": 1.5763011970574655e-06,
      "loss": 0.8071,
      "step": 2567
    },
    {
      "epoch": 0.7486334815246702,
      "grad_norm": 26.105068878511958,
      "learning_rate": 1.572862888324801e-06,
      "loss": 0.7905,
      "step": 2568
    },
    {
      "epoch": 0.7489250054660739,
      "grad_norm": 46.155439335326285,
      "learning_rate": 1.5694276335606707e-06,
      "loss": 0.8682,
      "step": 2569
    },
    {
      "epoch": 0.7492165294074776,
      "grad_norm": 41.59669875464882,
      "learning_rate": 1.5659954358262724e-06,
      "loss": 0.7087,
      "step": 2570
    },
    {
      "epoch": 0.7495080533488813,
      "grad_norm": 15.069528398393242,
      "learning_rate": 1.562566298180087e-06,
      "loss": 0.7944,
      "step": 2571
    },
    {
      "epoch": 0.749799577290285,
      "grad_norm": 24.12140997791283,
      "learning_rate": 1.5591402236778647e-06,
      "loss": 0.744,
      "step": 2572
    },
    {
      "epoch": 0.7500911012316887,
      "grad_norm": 41.89180347828711,
      "learning_rate": 1.555717215372628e-06,
      "loss": 0.7211,
      "step": 2573
    },
    {
      "epoch": 0.7503826251730923,
      "grad_norm": 20.09953693708435,
      "learning_rate": 1.5522972763146653e-06,
      "loss": 0.7795,
      "step": 2574
    },
    {
      "epoch": 0.750674149114496,
      "grad_norm": 49.02354258930029,
      "learning_rate": 1.5488804095515335e-06,
      "loss": 0.6885,
      "step": 2575
    },
    {
      "epoch": 0.7509656730558997,
      "grad_norm": 25.12693467843774,
      "learning_rate": 1.5454666181280437e-06,
      "loss": 0.7079,
      "step": 2576
    },
    {
      "epoch": 0.7512571969973034,
      "grad_norm": 67.43385633519532,
      "learning_rate": 1.5420559050862778e-06,
      "loss": 0.587,
      "step": 2577
    },
    {
      "epoch": 0.7515487209387071,
      "grad_norm": 27.042337045193094,
      "learning_rate": 1.5386482734655633e-06,
      "loss": 0.8337,
      "step": 2578
    },
    {
      "epoch": 0.7518402448801108,
      "grad_norm": 21.611750841537926,
      "learning_rate": 1.5352437263024928e-06,
      "loss": 0.6072,
      "step": 2579
    },
    {
      "epoch": 0.7521317688215144,
      "grad_norm": 33.99314859016171,
      "learning_rate": 1.5318422666308997e-06,
      "loss": 0.7172,
      "step": 2580
    },
    {
      "epoch": 0.7524232927629182,
      "grad_norm": 55.695332938566786,
      "learning_rate": 1.5284438974818732e-06,
      "loss": 0.7117,
      "step": 2581
    },
    {
      "epoch": 0.7527148167043218,
      "grad_norm": 21.004024716770665,
      "learning_rate": 1.5250486218837458e-06,
      "loss": 0.7637,
      "step": 2582
    },
    {
      "epoch": 0.7530063406457256,
      "grad_norm": 45.96142847422853,
      "learning_rate": 1.5216564428620957e-06,
      "loss": 0.7749,
      "step": 2583
    },
    {
      "epoch": 0.7532978645871292,
      "grad_norm": 28.525228044726582,
      "learning_rate": 1.5182673634397365e-06,
      "loss": 0.803,
      "step": 2584
    },
    {
      "epoch": 0.7535893885285329,
      "grad_norm": 28.080432327653796,
      "learning_rate": 1.5148813866367252e-06,
      "loss": 0.804,
      "step": 2585
    },
    {
      "epoch": 0.7538809124699366,
      "grad_norm": 19.035118765892378,
      "learning_rate": 1.5114985154703505e-06,
      "loss": 0.6462,
      "step": 2586
    },
    {
      "epoch": 0.7541724364113402,
      "grad_norm": 64.21249937616153,
      "learning_rate": 1.508118752955136e-06,
      "loss": 0.8029,
      "step": 2587
    },
    {
      "epoch": 0.754463960352744,
      "grad_norm": 33.47488895770595,
      "learning_rate": 1.5047421021028353e-06,
      "loss": 0.8621,
      "step": 2588
    },
    {
      "epoch": 0.7547554842941476,
      "grad_norm": 18.371526244281675,
      "learning_rate": 1.5013685659224237e-06,
      "loss": 0.7805,
      "step": 2589
    },
    {
      "epoch": 0.7550470082355514,
      "grad_norm": 33.0164825211254,
      "learning_rate": 1.4979981474201106e-06,
      "loss": 0.8203,
      "step": 2590
    },
    {
      "epoch": 0.755338532176955,
      "grad_norm": 35.72912440506265,
      "learning_rate": 1.494630849599316e-06,
      "loss": 0.5948,
      "step": 2591
    },
    {
      "epoch": 0.7556300561183588,
      "grad_norm": 27.889328638154357,
      "learning_rate": 1.4912666754606914e-06,
      "loss": 0.756,
      "step": 2592
    },
    {
      "epoch": 0.7559215800597624,
      "grad_norm": 25.99859563007984,
      "learning_rate": 1.4879056280020927e-06,
      "loss": 0.6973,
      "step": 2593
    },
    {
      "epoch": 0.7562131040011661,
      "grad_norm": 31.636229771521137,
      "learning_rate": 1.4845477102185974e-06,
      "loss": 0.7046,
      "step": 2594
    },
    {
      "epoch": 0.7565046279425698,
      "grad_norm": 26.325877327295068,
      "learning_rate": 1.4811929251024914e-06,
      "loss": 0.6503,
      "step": 2595
    },
    {
      "epoch": 0.7567961518839734,
      "grad_norm": 40.98920982280217,
      "learning_rate": 1.4778412756432709e-06,
      "loss": 0.7335,
      "step": 2596
    },
    {
      "epoch": 0.7570876758253772,
      "grad_norm": 24.4187916923671,
      "learning_rate": 1.474492764827632e-06,
      "loss": 0.7125,
      "step": 2597
    },
    {
      "epoch": 0.7573791997667808,
      "grad_norm": 19.537078794357992,
      "learning_rate": 1.471147395639484e-06,
      "loss": 0.6216,
      "step": 2598
    },
    {
      "epoch": 0.7576707237081846,
      "grad_norm": 23.366341378552562,
      "learning_rate": 1.467805171059926e-06,
      "loss": 0.5434,
      "step": 2599
    },
    {
      "epoch": 0.7579622476495882,
      "grad_norm": 29.386725454410964,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.7964,
      "step": 2600
    },
    {
      "epoch": 0.7579622476495882,
      "eval_loss": 1.1159924268722534,
      "eval_runtime": 2329.1792,
      "eval_samples_per_second": 7.415,
      "eval_steps_per_second": 0.412,
      "step": 2600
    },
    {
      "epoch": 0.7582537715909919,
      "grad_norm": 18.259142470259675,
      "learning_rate": 1.4611301676369899e-06,
      "loss": 0.7811,
      "step": 2601
    },
    {
      "epoch": 0.7585452955323956,
      "grad_norm": 34.50370458483662,
      "learning_rate": 1.457797394741798e-06,
      "loss": 0.8121,
      "step": 2602
    },
    {
      "epoch": 0.7588368194737993,
      "grad_norm": 56.41301509124602,
      "learning_rate": 1.4544677783515665e-06,
      "loss": 0.7792,
      "step": 2603
    },
    {
      "epoch": 0.759128343415203,
      "grad_norm": 28.161891902084196,
      "learning_rate": 1.451141321433358e-06,
      "loss": 0.7513,
      "step": 2604
    },
    {
      "epoch": 0.7594198673566067,
      "grad_norm": 29.8137090560073,
      "learning_rate": 1.4478180269514296e-06,
      "loss": 0.7822,
      "step": 2605
    },
    {
      "epoch": 0.7597113912980104,
      "grad_norm": 41.68380345065244,
      "learning_rate": 1.4444978978672103e-06,
      "loss": 0.7865,
      "step": 2606
    },
    {
      "epoch": 0.760002915239414,
      "grad_norm": 34.216611780022305,
      "learning_rate": 1.4411809371393133e-06,
      "loss": 0.6746,
      "step": 2607
    },
    {
      "epoch": 0.7602944391808177,
      "grad_norm": 34.48506118932083,
      "learning_rate": 1.4378671477235268e-06,
      "loss": 0.6755,
      "step": 2608
    },
    {
      "epoch": 0.7605859631222214,
      "grad_norm": 34.3018649773214,
      "learning_rate": 1.434556532572816e-06,
      "loss": 0.6637,
      "step": 2609
    },
    {
      "epoch": 0.7608774870636251,
      "grad_norm": 25.983098992477203,
      "learning_rate": 1.431249094637311e-06,
      "loss": 0.8344,
      "step": 2610
    },
    {
      "epoch": 0.7611690110050288,
      "grad_norm": 25.120408586448008,
      "learning_rate": 1.4279448368643195e-06,
      "loss": 0.5682,
      "step": 2611
    },
    {
      "epoch": 0.7614605349464325,
      "grad_norm": 32.09370330949585,
      "learning_rate": 1.4246437621983057e-06,
      "loss": 0.7443,
      "step": 2612
    },
    {
      "epoch": 0.7617520588878361,
      "grad_norm": 24.91481567435048,
      "learning_rate": 1.4213458735809072e-06,
      "loss": 0.7696,
      "step": 2613
    },
    {
      "epoch": 0.7620435828292399,
      "grad_norm": 25.07042628595904,
      "learning_rate": 1.418051173950914e-06,
      "loss": 0.6641,
      "step": 2614
    },
    {
      "epoch": 0.7623351067706435,
      "grad_norm": 38.99289734728756,
      "learning_rate": 1.4147596662442786e-06,
      "loss": 0.8629,
      "step": 2615
    },
    {
      "epoch": 0.7626266307120473,
      "grad_norm": 24.53030148875461,
      "learning_rate": 1.4114713533941082e-06,
      "loss": 0.7389,
      "step": 2616
    },
    {
      "epoch": 0.7629181546534509,
      "grad_norm": 34.14358284444594,
      "learning_rate": 1.4081862383306638e-06,
      "loss": 0.7208,
      "step": 2617
    },
    {
      "epoch": 0.7632096785948546,
      "grad_norm": 16.654282237338617,
      "learning_rate": 1.4049043239813575e-06,
      "loss": 0.7313,
      "step": 2618
    },
    {
      "epoch": 0.7635012025362583,
      "grad_norm": 41.790522949435925,
      "learning_rate": 1.4016256132707445e-06,
      "loss": 0.7146,
      "step": 2619
    },
    {
      "epoch": 0.7637927264776619,
      "grad_norm": 25.439251934644307,
      "learning_rate": 1.3983501091205298e-06,
      "loss": 0.6578,
      "step": 2620
    },
    {
      "epoch": 0.7640842504190657,
      "grad_norm": 23.379191345867664,
      "learning_rate": 1.3950778144495608e-06,
      "loss": 0.7068,
      "step": 2621
    },
    {
      "epoch": 0.7643757743604693,
      "grad_norm": 59.43338454615795,
      "learning_rate": 1.3918087321738244e-06,
      "loss": 0.5745,
      "step": 2622
    },
    {
      "epoch": 0.7646672983018731,
      "grad_norm": 25.03305218318833,
      "learning_rate": 1.3885428652064398e-06,
      "loss": 0.6776,
      "step": 2623
    },
    {
      "epoch": 0.7649588222432767,
      "grad_norm": 23.506052450036094,
      "learning_rate": 1.3852802164576717e-06,
      "loss": 0.59,
      "step": 2624
    },
    {
      "epoch": 0.7652503461846805,
      "grad_norm": 44.65122115235825,
      "learning_rate": 1.3820207888349052e-06,
      "loss": 0.8145,
      "step": 2625
    },
    {
      "epoch": 0.7655418701260841,
      "grad_norm": 96.96470711780593,
      "learning_rate": 1.3787645852426663e-06,
      "loss": 0.7342,
      "step": 2626
    },
    {
      "epoch": 0.7658333940674877,
      "grad_norm": 24.135310115127478,
      "learning_rate": 1.3755116085825975e-06,
      "loss": 0.7446,
      "step": 2627
    },
    {
      "epoch": 0.7661249180088915,
      "grad_norm": 23.869479419524517,
      "learning_rate": 1.3722618617534727e-06,
      "loss": 0.6313,
      "step": 2628
    },
    {
      "epoch": 0.7664164419502951,
      "grad_norm": 39.111384375817494,
      "learning_rate": 1.3690153476511852e-06,
      "loss": 0.9456,
      "step": 2629
    },
    {
      "epoch": 0.7667079658916989,
      "grad_norm": 16.433756827854104,
      "learning_rate": 1.3657720691687481e-06,
      "loss": 0.6656,
      "step": 2630
    },
    {
      "epoch": 0.7669994898331025,
      "grad_norm": 69.58574557517233,
      "learning_rate": 1.3625320291962907e-06,
      "loss": 0.7473,
      "step": 2631
    },
    {
      "epoch": 0.7672910137745063,
      "grad_norm": 29.73817013393585,
      "learning_rate": 1.3592952306210589e-06,
      "loss": 0.6601,
      "step": 2632
    },
    {
      "epoch": 0.7675825377159099,
      "grad_norm": 33.389448252783026,
      "learning_rate": 1.3560616763274042e-06,
      "loss": 0.6894,
      "step": 2633
    },
    {
      "epoch": 0.7678740616573136,
      "grad_norm": 31.257665563968114,
      "learning_rate": 1.3528313691967926e-06,
      "loss": 0.6406,
      "step": 2634
    },
    {
      "epoch": 0.7681655855987173,
      "grad_norm": 62.12531303302557,
      "learning_rate": 1.3496043121077946e-06,
      "loss": 0.6502,
      "step": 2635
    },
    {
      "epoch": 0.768457109540121,
      "grad_norm": 24.95771057875098,
      "learning_rate": 1.3463805079360854e-06,
      "loss": 0.7743,
      "step": 2636
    },
    {
      "epoch": 0.7687486334815247,
      "grad_norm": 63.34642607119198,
      "learning_rate": 1.3431599595544415e-06,
      "loss": 0.6208,
      "step": 2637
    },
    {
      "epoch": 0.7690401574229283,
      "grad_norm": 31.746080402078118,
      "learning_rate": 1.3399426698327329e-06,
      "loss": 0.7005,
      "step": 2638
    },
    {
      "epoch": 0.769331681364332,
      "grad_norm": 23.730791162234386,
      "learning_rate": 1.3367286416379366e-06,
      "loss": 0.6361,
      "step": 2639
    },
    {
      "epoch": 0.7696232053057357,
      "grad_norm": 19.487598693619077,
      "learning_rate": 1.3335178778341123e-06,
      "loss": 0.6781,
      "step": 2640
    },
    {
      "epoch": 0.7699147292471394,
      "grad_norm": 37.36313330591283,
      "learning_rate": 1.3303103812824176e-06,
      "loss": 0.65,
      "step": 2641
    },
    {
      "epoch": 0.7702062531885431,
      "grad_norm": 28.761541143603587,
      "learning_rate": 1.3271061548410947e-06,
      "loss": 0.67,
      "step": 2642
    },
    {
      "epoch": 0.7704977771299468,
      "grad_norm": 40.80003436114947,
      "learning_rate": 1.3239052013654747e-06,
      "loss": 0.7805,
      "step": 2643
    },
    {
      "epoch": 0.7707893010713505,
      "grad_norm": 35.299175623755936,
      "learning_rate": 1.3207075237079702e-06,
      "loss": 0.9156,
      "step": 2644
    },
    {
      "epoch": 0.7710808250127542,
      "grad_norm": 27.56735256852195,
      "learning_rate": 1.3175131247180773e-06,
      "loss": 0.8308,
      "step": 2645
    },
    {
      "epoch": 0.7713723489541578,
      "grad_norm": 33.83381551001268,
      "learning_rate": 1.3143220072423647e-06,
      "loss": 0.6179,
      "step": 2646
    },
    {
      "epoch": 0.7716638728955616,
      "grad_norm": 42.11094051765864,
      "learning_rate": 1.3111341741244826e-06,
      "loss": 0.7241,
      "step": 2647
    },
    {
      "epoch": 0.7719553968369652,
      "grad_norm": 31.296840880783744,
      "learning_rate": 1.307949628205153e-06,
      "loss": 0.7781,
      "step": 2648
    },
    {
      "epoch": 0.7722469207783689,
      "grad_norm": 21.85678204846229,
      "learning_rate": 1.3047683723221667e-06,
      "loss": 0.7747,
      "step": 2649
    },
    {
      "epoch": 0.7725384447197726,
      "grad_norm": 165.92447932046616,
      "learning_rate": 1.301590409310387e-06,
      "loss": 0.7495,
      "step": 2650
    },
    {
      "epoch": 0.7725384447197726,
      "eval_loss": 1.110398292541504,
      "eval_runtime": 2359.1348,
      "eval_samples_per_second": 7.32,
      "eval_steps_per_second": 0.407,
      "step": 2650
    },
    {
      "epoch": 0.7728299686611763,
      "grad_norm": 30.428082984951825,
      "learning_rate": 1.2984157420017346e-06,
      "loss": 0.6539,
      "step": 2651
    },
    {
      "epoch": 0.77312149260258,
      "grad_norm": 41.60167942431836,
      "learning_rate": 1.2952443732252058e-06,
      "loss": 0.7341,
      "step": 2652
    },
    {
      "epoch": 0.7734130165439836,
      "grad_norm": 19.31266199233225,
      "learning_rate": 1.2920763058068448e-06,
      "loss": 0.6168,
      "step": 2653
    },
    {
      "epoch": 0.7737045404853874,
      "grad_norm": 15.435941813162806,
      "learning_rate": 1.2889115425697612e-06,
      "loss": 0.5663,
      "step": 2654
    },
    {
      "epoch": 0.773996064426791,
      "grad_norm": 103.43760934412798,
      "learning_rate": 1.28575008633412e-06,
      "loss": 0.7573,
      "step": 2655
    },
    {
      "epoch": 0.7742875883681948,
      "grad_norm": 27.388518798526448,
      "learning_rate": 1.282591939917136e-06,
      "loss": 0.7308,
      "step": 2656
    },
    {
      "epoch": 0.7745791123095984,
      "grad_norm": 26.616098126135654,
      "learning_rate": 1.2794371061330779e-06,
      "loss": 0.6124,
      "step": 2657
    },
    {
      "epoch": 0.7748706362510022,
      "grad_norm": 25.507448471428464,
      "learning_rate": 1.2762855877932617e-06,
      "loss": 0.745,
      "step": 2658
    },
    {
      "epoch": 0.7751621601924058,
      "grad_norm": 24.370368530603148,
      "learning_rate": 1.2731373877060438e-06,
      "loss": 0.6422,
      "step": 2659
    },
    {
      "epoch": 0.7754536841338094,
      "grad_norm": 21.81157694920732,
      "learning_rate": 1.269992508676835e-06,
      "loss": 0.7199,
      "step": 2660
    },
    {
      "epoch": 0.7757452080752132,
      "grad_norm": 60.28682744185092,
      "learning_rate": 1.2668509535080747e-06,
      "loss": 0.6259,
      "step": 2661
    },
    {
      "epoch": 0.7760367320166168,
      "grad_norm": 22.39438408517215,
      "learning_rate": 1.2637127249992465e-06,
      "loss": 0.7247,
      "step": 2662
    },
    {
      "epoch": 0.7763282559580206,
      "grad_norm": 68.04270360819302,
      "learning_rate": 1.2605778259468693e-06,
      "loss": 0.8649,
      "step": 2663
    },
    {
      "epoch": 0.7766197798994242,
      "grad_norm": 17.855876323817316,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.6639,
      "step": 2664
    },
    {
      "epoch": 0.776911303840828,
      "grad_norm": 45.689901752176084,
      "learning_rate": 1.2543180273827044e-06,
      "loss": 0.7825,
      "step": 2665
    },
    {
      "epoch": 0.7772028277822316,
      "grad_norm": 31.73920968424702,
      "learning_rate": 1.2511931334491068e-06,
      "loss": 0.7282,
      "step": 2666
    },
    {
      "epoch": 0.7774943517236353,
      "grad_norm": 43.328968319197365,
      "learning_rate": 1.2480715801283382e-06,
      "loss": 0.7071,
      "step": 2667
    },
    {
      "epoch": 0.777785875665039,
      "grad_norm": 34.84202765897263,
      "learning_rate": 1.2449533702020578e-06,
      "loss": 0.6015,
      "step": 2668
    },
    {
      "epoch": 0.7780773996064427,
      "grad_norm": 32.55481851953895,
      "learning_rate": 1.241838506448944e-06,
      "loss": 0.6406,
      "step": 2669
    },
    {
      "epoch": 0.7783689235478464,
      "grad_norm": 41.709734293137686,
      "learning_rate": 1.238726991644696e-06,
      "loss": 0.7221,
      "step": 2670
    },
    {
      "epoch": 0.77866044748925,
      "grad_norm": 35.768438957955794,
      "learning_rate": 1.2356188285620262e-06,
      "loss": 0.8066,
      "step": 2671
    },
    {
      "epoch": 0.7789519714306538,
      "grad_norm": 32.57558466224305,
      "learning_rate": 1.232514019970658e-06,
      "loss": 0.6717,
      "step": 2672
    },
    {
      "epoch": 0.7792434953720574,
      "grad_norm": 31.60561095375762,
      "learning_rate": 1.2294125686373348e-06,
      "loss": 0.5919,
      "step": 2673
    },
    {
      "epoch": 0.7795350193134611,
      "grad_norm": 26.156526468546208,
      "learning_rate": 1.2263144773257967e-06,
      "loss": 0.8709,
      "step": 2674
    },
    {
      "epoch": 0.7798265432548648,
      "grad_norm": 22.205304315607428,
      "learning_rate": 1.2232197487967967e-06,
      "loss": 0.701,
      "step": 2675
    },
    {
      "epoch": 0.7801180671962685,
      "grad_norm": 24.265406221436283,
      "learning_rate": 1.2201283858080903e-06,
      "loss": 0.6284,
      "step": 2676
    },
    {
      "epoch": 0.7804095911376722,
      "grad_norm": 24.455610305183853,
      "learning_rate": 1.2170403911144325e-06,
      "loss": 0.6824,
      "step": 2677
    },
    {
      "epoch": 0.7807011150790759,
      "grad_norm": 25.810666726472434,
      "learning_rate": 1.2139557674675773e-06,
      "loss": 0.7199,
      "step": 2678
    },
    {
      "epoch": 0.7809926390204796,
      "grad_norm": 26.221743221946515,
      "learning_rate": 1.210874517616276e-06,
      "loss": 0.5907,
      "step": 2679
    },
    {
      "epoch": 0.7812841629618833,
      "grad_norm": 17.136880759064795,
      "learning_rate": 1.2077966443062706e-06,
      "loss": 0.576,
      "step": 2680
    },
    {
      "epoch": 0.7815756869032869,
      "grad_norm": 20.820007555388507,
      "learning_rate": 1.2047221502802963e-06,
      "loss": 0.6261,
      "step": 2681
    },
    {
      "epoch": 0.7818672108446906,
      "grad_norm": 31.80545203112631,
      "learning_rate": 1.2016510382780772e-06,
      "loss": 0.7508,
      "step": 2682
    },
    {
      "epoch": 0.7821587347860943,
      "grad_norm": 21.56095415649906,
      "learning_rate": 1.1985833110363226e-06,
      "loss": 0.6356,
      "step": 2683
    },
    {
      "epoch": 0.782450258727498,
      "grad_norm": 33.58905750708046,
      "learning_rate": 1.1955189712887272e-06,
      "loss": 0.6363,
      "step": 2684
    },
    {
      "epoch": 0.7827417826689017,
      "grad_norm": 23.257971542962945,
      "learning_rate": 1.192458021765962e-06,
      "loss": 0.5989,
      "step": 2685
    },
    {
      "epoch": 0.7830333066103053,
      "grad_norm": 20.690944293253214,
      "learning_rate": 1.189400465195687e-06,
      "loss": 0.8034,
      "step": 2686
    },
    {
      "epoch": 0.7833248305517091,
      "grad_norm": 33.69596368728512,
      "learning_rate": 1.186346304302527e-06,
      "loss": 0.7574,
      "step": 2687
    },
    {
      "epoch": 0.7836163544931127,
      "grad_norm": 18.516107251637532,
      "learning_rate": 1.183295541808089e-06,
      "loss": 0.6631,
      "step": 2688
    },
    {
      "epoch": 0.7839078784345165,
      "grad_norm": 31.437123172671214,
      "learning_rate": 1.1802481804309473e-06,
      "loss": 0.6707,
      "step": 2689
    },
    {
      "epoch": 0.7841994023759201,
      "grad_norm": 44.901149199476244,
      "learning_rate": 1.1772042228866493e-06,
      "loss": 0.7124,
      "step": 2690
    },
    {
      "epoch": 0.7844909263173238,
      "grad_norm": 40.21342481158566,
      "learning_rate": 1.1741636718877053e-06,
      "loss": 0.7005,
      "step": 2691
    },
    {
      "epoch": 0.7847824502587275,
      "grad_norm": 19.56595642119326,
      "learning_rate": 1.1711265301435937e-06,
      "loss": 0.6941,
      "step": 2692
    },
    {
      "epoch": 0.7850739742001311,
      "grad_norm": 42.88765378811451,
      "learning_rate": 1.1680928003607488e-06,
      "loss": 0.641,
      "step": 2693
    },
    {
      "epoch": 0.7853654981415349,
      "grad_norm": 23.607076831127067,
      "learning_rate": 1.165062485242574e-06,
      "loss": 0.843,
      "step": 2694
    },
    {
      "epoch": 0.7856570220829385,
      "grad_norm": 35.40318743779092,
      "learning_rate": 1.1620355874894195e-06,
      "loss": 0.5138,
      "step": 2695
    },
    {
      "epoch": 0.7859485460243423,
      "grad_norm": 23.453057074878263,
      "learning_rate": 1.159012109798598e-06,
      "loss": 0.9004,
      "step": 2696
    },
    {
      "epoch": 0.7862400699657459,
      "grad_norm": 27.02915593574601,
      "learning_rate": 1.15599205486437e-06,
      "loss": 0.678,
      "step": 2697
    },
    {
      "epoch": 0.7865315939071497,
      "grad_norm": 21.77568036738578,
      "learning_rate": 1.1529754253779486e-06,
      "loss": 0.7901,
      "step": 2698
    },
    {
      "epoch": 0.7868231178485533,
      "grad_norm": 20.795079767467097,
      "learning_rate": 1.1499622240274938e-06,
      "loss": 0.7551,
      "step": 2699
    },
    {
      "epoch": 0.787114641789957,
      "grad_norm": 107.65888328594365,
      "learning_rate": 1.1469524534981091e-06,
      "loss": 0.7217,
      "step": 2700
    },
    {
      "epoch": 0.787114641789957,
      "eval_loss": 1.1164857149124146,
      "eval_runtime": 2357.5612,
      "eval_samples_per_second": 7.325,
      "eval_steps_per_second": 0.407,
      "step": 2700
    },
    {
      "epoch": 0.7874061657313607,
      "grad_norm": 39.63207342861569,
      "learning_rate": 1.1439461164718418e-06,
      "loss": 0.859,
      "step": 2701
    },
    {
      "epoch": 0.7876976896727643,
      "grad_norm": 83.76840276786318,
      "learning_rate": 1.1409432156276805e-06,
      "loss": 0.7175,
      "step": 2702
    },
    {
      "epoch": 0.7879892136141681,
      "grad_norm": 24.199099782313226,
      "learning_rate": 1.1379437536415516e-06,
      "loss": 0.6938,
      "step": 2703
    },
    {
      "epoch": 0.7882807375555717,
      "grad_norm": 21.60418121529034,
      "learning_rate": 1.134947733186315e-06,
      "loss": 0.8161,
      "step": 2704
    },
    {
      "epoch": 0.7885722614969755,
      "grad_norm": 24.593544682115386,
      "learning_rate": 1.1319551569317683e-06,
      "loss": 0.6112,
      "step": 2705
    },
    {
      "epoch": 0.7888637854383791,
      "grad_norm": 17.48721514491563,
      "learning_rate": 1.1289660275446318e-06,
      "loss": 0.6563,
      "step": 2706
    },
    {
      "epoch": 0.7891553093797828,
      "grad_norm": 44.50858353024652,
      "learning_rate": 1.125980347688565e-06,
      "loss": 0.7132,
      "step": 2707
    },
    {
      "epoch": 0.7894468333211865,
      "grad_norm": 29.218712852274034,
      "learning_rate": 1.1229981200241424e-06,
      "loss": 0.7571,
      "step": 2708
    },
    {
      "epoch": 0.7897383572625902,
      "grad_norm": 25.40224534943177,
      "learning_rate": 1.1200193472088733e-06,
      "loss": 0.7424,
      "step": 2709
    },
    {
      "epoch": 0.7900298812039939,
      "grad_norm": 20.13777990608553,
      "learning_rate": 1.1170440318971788e-06,
      "loss": 0.7691,
      "step": 2710
    },
    {
      "epoch": 0.7903214051453976,
      "grad_norm": 18.902406297853535,
      "learning_rate": 1.114072176740404e-06,
      "loss": 0.7008,
      "step": 2711
    },
    {
      "epoch": 0.7906129290868013,
      "grad_norm": 24.50010780141035,
      "learning_rate": 1.1111037843868095e-06,
      "loss": 0.7651,
      "step": 2712
    },
    {
      "epoch": 0.7909044530282049,
      "grad_norm": 29.42443922871724,
      "learning_rate": 1.1081388574815711e-06,
      "loss": 0.7857,
      "step": 2713
    },
    {
      "epoch": 0.7911959769696086,
      "grad_norm": 39.09334515172981,
      "learning_rate": 1.1051773986667735e-06,
      "loss": 0.7608,
      "step": 2714
    },
    {
      "epoch": 0.7914875009110123,
      "grad_norm": 23.381306106717687,
      "learning_rate": 1.1022194105814137e-06,
      "loss": 0.7003,
      "step": 2715
    },
    {
      "epoch": 0.791779024852416,
      "grad_norm": 26.39632742586717,
      "learning_rate": 1.0992648958613961e-06,
      "loss": 0.6122,
      "step": 2716
    },
    {
      "epoch": 0.7920705487938197,
      "grad_norm": 38.93011278682053,
      "learning_rate": 1.0963138571395277e-06,
      "loss": 0.5813,
      "step": 2717
    },
    {
      "epoch": 0.7923620727352234,
      "grad_norm": 23.334888022621357,
      "learning_rate": 1.0933662970455217e-06,
      "loss": 0.8581,
      "step": 2718
    },
    {
      "epoch": 0.792653596676627,
      "grad_norm": 36.66154710074728,
      "learning_rate": 1.0904222182059843e-06,
      "loss": 0.786,
      "step": 2719
    },
    {
      "epoch": 0.7929451206180308,
      "grad_norm": 46.225691006394264,
      "learning_rate": 1.0874816232444297e-06,
      "loss": 0.7555,
      "step": 2720
    },
    {
      "epoch": 0.7932366445594344,
      "grad_norm": 23.74837448332179,
      "learning_rate": 1.0845445147812573e-06,
      "loss": 0.8611,
      "step": 2721
    },
    {
      "epoch": 0.7935281685008382,
      "grad_norm": 19.49959744713486,
      "learning_rate": 1.081610895433769e-06,
      "loss": 0.8116,
      "step": 2722
    },
    {
      "epoch": 0.7938196924422418,
      "grad_norm": 26.91872685154897,
      "learning_rate": 1.0786807678161499e-06,
      "loss": 0.7796,
      "step": 2723
    },
    {
      "epoch": 0.7941112163836455,
      "grad_norm": 23.168519106258966,
      "learning_rate": 1.0757541345394768e-06,
      "loss": 0.6081,
      "step": 2724
    },
    {
      "epoch": 0.7944027403250492,
      "grad_norm": 24.888954645955835,
      "learning_rate": 1.072830998211713e-06,
      "loss": 0.6826,
      "step": 2725
    },
    {
      "epoch": 0.7946942642664528,
      "grad_norm": 28.619118547206973,
      "learning_rate": 1.0699113614377065e-06,
      "loss": 0.7037,
      "step": 2726
    },
    {
      "epoch": 0.7949857882078566,
      "grad_norm": 20.12534421040863,
      "learning_rate": 1.0669952268191825e-06,
      "loss": 0.8293,
      "step": 2727
    },
    {
      "epoch": 0.7952773121492602,
      "grad_norm": 33.42000519003641,
      "learning_rate": 1.0640825969547498e-06,
      "loss": 0.6935,
      "step": 2728
    },
    {
      "epoch": 0.795568836090664,
      "grad_norm": 19.17082232439525,
      "learning_rate": 1.0611734744398922e-06,
      "loss": 0.9118,
      "step": 2729
    },
    {
      "epoch": 0.7958603600320676,
      "grad_norm": 62.79040912046935,
      "learning_rate": 1.058267861866969e-06,
      "loss": 0.6472,
      "step": 2730
    },
    {
      "epoch": 0.7961518839734714,
      "grad_norm": 38.33887026730426,
      "learning_rate": 1.0553657618252116e-06,
      "loss": 0.671,
      "step": 2731
    },
    {
      "epoch": 0.796443407914875,
      "grad_norm": 22.72465550517727,
      "learning_rate": 1.0524671769007177e-06,
      "loss": 0.8728,
      "step": 2732
    },
    {
      "epoch": 0.7967349318562787,
      "grad_norm": 27.810257979459188,
      "learning_rate": 1.0495721096764604e-06,
      "loss": 0.6213,
      "step": 2733
    },
    {
      "epoch": 0.7970264557976824,
      "grad_norm": 27.98887445215835,
      "learning_rate": 1.0466805627322685e-06,
      "loss": 0.9432,
      "step": 2734
    },
    {
      "epoch": 0.797317979739086,
      "grad_norm": 50.67429770230281,
      "learning_rate": 1.0437925386448444e-06,
      "loss": 0.8606,
      "step": 2735
    },
    {
      "epoch": 0.7976095036804898,
      "grad_norm": 51.45539744645453,
      "learning_rate": 1.0409080399877413e-06,
      "loss": 0.6814,
      "step": 2736
    },
    {
      "epoch": 0.7979010276218934,
      "grad_norm": 33.23613693355536,
      "learning_rate": 1.0380270693313772e-06,
      "loss": 0.6951,
      "step": 2737
    },
    {
      "epoch": 0.7981925515632972,
      "grad_norm": 30.3860429484072,
      "learning_rate": 1.035149629243023e-06,
      "loss": 0.7771,
      "step": 2738
    },
    {
      "epoch": 0.7984840755047008,
      "grad_norm": 39.66407017269368,
      "learning_rate": 1.032275722286807e-06,
      "loss": 0.692,
      "step": 2739
    },
    {
      "epoch": 0.7987755994461045,
      "grad_norm": 32.04987112102548,
      "learning_rate": 1.0294053510237028e-06,
      "loss": 0.724,
      "step": 2740
    },
    {
      "epoch": 0.7990671233875082,
      "grad_norm": 34.990225867111384,
      "learning_rate": 1.0265385180115423e-06,
      "loss": 0.7034,
      "step": 2741
    },
    {
      "epoch": 0.7993586473289119,
      "grad_norm": 30.178345274406137,
      "learning_rate": 1.0236752258049954e-06,
      "loss": 0.6788,
      "step": 2742
    },
    {
      "epoch": 0.7996501712703156,
      "grad_norm": 24.45105867213991,
      "learning_rate": 1.0208154769555828e-06,
      "loss": 0.7851,
      "step": 2743
    },
    {
      "epoch": 0.7999416952117193,
      "grad_norm": 34.11780907212069,
      "learning_rate": 1.017959274011665e-06,
      "loss": 0.7313,
      "step": 2744
    },
    {
      "epoch": 0.800233219153123,
      "grad_norm": 20.013686610631957,
      "learning_rate": 1.0151066195184435e-06,
      "loss": 0.7757,
      "step": 2745
    },
    {
      "epoch": 0.8005247430945266,
      "grad_norm": 19.94654379168293,
      "learning_rate": 1.0122575160179582e-06,
      "loss": 0.7112,
      "step": 2746
    },
    {
      "epoch": 0.8008162670359303,
      "grad_norm": 31.836010114188678,
      "learning_rate": 1.009411966049081e-06,
      "loss": 0.7454,
      "step": 2747
    },
    {
      "epoch": 0.801107790977334,
      "grad_norm": 32.47954668874927,
      "learning_rate": 1.0065699721475253e-06,
      "loss": 0.7177,
      "step": 2748
    },
    {
      "epoch": 0.8013993149187377,
      "grad_norm": 19.816755129934613,
      "learning_rate": 1.0037315368458266e-06,
      "loss": 0.6242,
      "step": 2749
    },
    {
      "epoch": 0.8016908388601414,
      "grad_norm": 37.03297799924849,
      "learning_rate": 1.0008966626733541e-06,
      "loss": 0.7871,
      "step": 2750
    },
    {
      "epoch": 0.8016908388601414,
      "eval_loss": 1.1137112379074097,
      "eval_runtime": 2407.0387,
      "eval_samples_per_second": 7.175,
      "eval_steps_per_second": 0.399,
      "step": 2750
    },
    {
      "epoch": 0.8019823628015451,
      "grad_norm": 26.254110831027013,
      "learning_rate": 9.980653521563033e-07,
      "loss": 0.8284,
      "step": 2751
    },
    {
      "epoch": 0.8022738867429487,
      "grad_norm": 29.11204679596526,
      "learning_rate": 9.95237607817694e-07,
      "loss": 0.7968,
      "step": 2752
    },
    {
      "epoch": 0.8025654106843525,
      "grad_norm": 33.00412742404926,
      "learning_rate": 9.92413432177365e-07,
      "loss": 0.8407,
      "step": 2753
    },
    {
      "epoch": 0.8028569346257561,
      "grad_norm": 25.22698825428694,
      "learning_rate": 9.895928277519822e-07,
      "loss": 0.6948,
      "step": 2754
    },
    {
      "epoch": 0.8031484585671598,
      "grad_norm": 28.171017657047724,
      "learning_rate": 9.86775797055019e-07,
      "loss": 0.6995,
      "step": 2755
    },
    {
      "epoch": 0.8034399825085635,
      "grad_norm": 25.747585703612287,
      "learning_rate": 9.83962342596776e-07,
      "loss": 0.6947,
      "step": 2756
    },
    {
      "epoch": 0.8037315064499672,
      "grad_norm": 35.83695985693653,
      "learning_rate": 9.81152466884356e-07,
      "loss": 0.8061,
      "step": 2757
    },
    {
      "epoch": 0.8040230303913709,
      "grad_norm": 51.754664659672116,
      "learning_rate": 9.783461724216793e-07,
      "loss": 0.6695,
      "step": 2758
    },
    {
      "epoch": 0.8043145543327745,
      "grad_norm": 24.921817488607594,
      "learning_rate": 9.755434617094727e-07,
      "loss": 0.8049,
      "step": 2759
    },
    {
      "epoch": 0.8046060782741783,
      "grad_norm": 190.15170040974132,
      "learning_rate": 9.7274433724527e-07,
      "loss": 0.7014,
      "step": 2760
    },
    {
      "epoch": 0.8048976022155819,
      "grad_norm": 20.171897074241993,
      "learning_rate": 9.699488015234104e-07,
      "loss": 0.7946,
      "step": 2761
    },
    {
      "epoch": 0.8051891261569857,
      "grad_norm": 21.680246768183043,
      "learning_rate": 9.671568570350321e-07,
      "loss": 0.9709,
      "step": 2762
    },
    {
      "epoch": 0.8054806500983893,
      "grad_norm": 20.814160722313865,
      "learning_rate": 9.643685062680746e-07,
      "loss": 0.652,
      "step": 2763
    },
    {
      "epoch": 0.8057721740397931,
      "grad_norm": 27.51551636575316,
      "learning_rate": 9.615837517072758e-07,
      "loss": 0.7981,
      "step": 2764
    },
    {
      "epoch": 0.8060636979811967,
      "grad_norm": 50.73386988088008,
      "learning_rate": 9.588025958341701e-07,
      "loss": 0.6627,
      "step": 2765
    },
    {
      "epoch": 0.8063552219226003,
      "grad_norm": 32.63032022442447,
      "learning_rate": 9.560250411270794e-07,
      "loss": 0.5894,
      "step": 2766
    },
    {
      "epoch": 0.8066467458640041,
      "grad_norm": 18.342532004799185,
      "learning_rate": 9.532510900611258e-07,
      "loss": 0.7153,
      "step": 2767
    },
    {
      "epoch": 0.8069382698054077,
      "grad_norm": 25.620056839920537,
      "learning_rate": 9.504807451082088e-07,
      "loss": 0.5957,
      "step": 2768
    },
    {
      "epoch": 0.8072297937468115,
      "grad_norm": 30.918417566974917,
      "learning_rate": 9.477140087370268e-07,
      "loss": 0.7173,
      "step": 2769
    },
    {
      "epoch": 0.8075213176882151,
      "grad_norm": 29.501081209572938,
      "learning_rate": 9.449508834130517e-07,
      "loss": 0.7086,
      "step": 2770
    },
    {
      "epoch": 0.8078128416296189,
      "grad_norm": 23.863511739690644,
      "learning_rate": 9.421913715985431e-07,
      "loss": 0.7011,
      "step": 2771
    },
    {
      "epoch": 0.8081043655710225,
      "grad_norm": 37.015756416493865,
      "learning_rate": 9.394354757525404e-07,
      "loss": 0.7576,
      "step": 2772
    },
    {
      "epoch": 0.8083958895124262,
      "grad_norm": 22.799748986368307,
      "learning_rate": 9.366831983308594e-07,
      "loss": 0.6761,
      "step": 2773
    },
    {
      "epoch": 0.8086874134538299,
      "grad_norm": 26.018432725284416,
      "learning_rate": 9.339345417860918e-07,
      "loss": 0.7568,
      "step": 2774
    },
    {
      "epoch": 0.8089789373952336,
      "grad_norm": 44.04809515673153,
      "learning_rate": 9.311895085676049e-07,
      "loss": 0.7881,
      "step": 2775
    },
    {
      "epoch": 0.8092704613366373,
      "grad_norm": 19.333187512412767,
      "learning_rate": 9.284481011215318e-07,
      "loss": 0.5856,
      "step": 2776
    },
    {
      "epoch": 0.8095619852780409,
      "grad_norm": 22.58674891034968,
      "learning_rate": 9.257103218907793e-07,
      "loss": 0.6392,
      "step": 2777
    },
    {
      "epoch": 0.8098535092194447,
      "grad_norm": 50.65617256754764,
      "learning_rate": 9.229761733150205e-07,
      "loss": 0.8845,
      "step": 2778
    },
    {
      "epoch": 0.8101450331608483,
      "grad_norm": 23.538707062665477,
      "learning_rate": 9.202456578306912e-07,
      "loss": 0.7249,
      "step": 2779
    },
    {
      "epoch": 0.810436557102252,
      "grad_norm": 35.03879744370955,
      "learning_rate": 9.175187778709937e-07,
      "loss": 0.6174,
      "step": 2780
    },
    {
      "epoch": 0.8107280810436557,
      "grad_norm": 28.600148188986932,
      "learning_rate": 9.147955358658827e-07,
      "loss": 0.8676,
      "step": 2781
    },
    {
      "epoch": 0.8110196049850594,
      "grad_norm": 26.7478118205036,
      "learning_rate": 9.120759342420821e-07,
      "loss": 0.7138,
      "step": 2782
    },
    {
      "epoch": 0.8113111289264631,
      "grad_norm": 26.397309036750034,
      "learning_rate": 9.093599754230619e-07,
      "loss": 0.6533,
      "step": 2783
    },
    {
      "epoch": 0.8116026528678668,
      "grad_norm": 85.83649296720898,
      "learning_rate": 9.066476618290515e-07,
      "loss": 0.8454,
      "step": 2784
    },
    {
      "epoch": 0.8118941768092705,
      "grad_norm": 36.139428328474544,
      "learning_rate": 9.039389958770295e-07,
      "loss": 0.7033,
      "step": 2785
    },
    {
      "epoch": 0.8121857007506742,
      "grad_norm": 46.16871845185345,
      "learning_rate": 9.012339799807263e-07,
      "loss": 0.7746,
      "step": 2786
    },
    {
      "epoch": 0.8124772246920778,
      "grad_norm": 24.594763490854504,
      "learning_rate": 8.985326165506175e-07,
      "loss": 0.6764,
      "step": 2787
    },
    {
      "epoch": 0.8127687486334815,
      "grad_norm": 50.771819504954635,
      "learning_rate": 8.95834907993926e-07,
      "loss": 0.7989,
      "step": 2788
    },
    {
      "epoch": 0.8130602725748852,
      "grad_norm": 30.258739972727597,
      "learning_rate": 8.931408567146138e-07,
      "loss": 0.7741,
      "step": 2789
    },
    {
      "epoch": 0.8133517965162889,
      "grad_norm": 36.73334270670326,
      "learning_rate": 8.904504651133905e-07,
      "loss": 0.7706,
      "step": 2790
    },
    {
      "epoch": 0.8136433204576926,
      "grad_norm": 43.181439102487744,
      "learning_rate": 8.877637355876978e-07,
      "loss": 0.7135,
      "step": 2791
    },
    {
      "epoch": 0.8139348443990962,
      "grad_norm": 20.827182502681367,
      "learning_rate": 8.850806705317183e-07,
      "loss": 0.5405,
      "step": 2792
    },
    {
      "epoch": 0.8142263683405,
      "grad_norm": 25.42516135204259,
      "learning_rate": 8.824012723363667e-07,
      "loss": 0.754,
      "step": 2793
    },
    {
      "epoch": 0.8145178922819036,
      "grad_norm": 54.2243710023909,
      "learning_rate": 8.797255433892926e-07,
      "loss": 0.6526,
      "step": 2794
    },
    {
      "epoch": 0.8148094162233074,
      "grad_norm": 42.19917168383912,
      "learning_rate": 8.770534860748747e-07,
      "loss": 0.6292,
      "step": 2795
    },
    {
      "epoch": 0.815100940164711,
      "grad_norm": 748.0036246337894,
      "learning_rate": 8.743851027742172e-07,
      "loss": 0.6925,
      "step": 2796
    },
    {
      "epoch": 0.8153924641061148,
      "grad_norm": 25.689005148585263,
      "learning_rate": 8.717203958651549e-07,
      "loss": 0.6726,
      "step": 2797
    },
    {
      "epoch": 0.8156839880475184,
      "grad_norm": 18.511545042159156,
      "learning_rate": 8.690593677222431e-07,
      "loss": 0.5864,
      "step": 2798
    },
    {
      "epoch": 0.815975511988922,
      "grad_norm": 26.231465402262533,
      "learning_rate": 8.664020207167617e-07,
      "loss": 0.7314,
      "step": 2799
    },
    {
      "epoch": 0.8162670359303258,
      "grad_norm": 34.73237571242994,
      "learning_rate": 8.637483572167077e-07,
      "loss": 0.8414,
      "step": 2800
    },
    {
      "epoch": 0.8162670359303258,
      "eval_loss": 1.1124855279922485,
      "eval_runtime": 2331.0996,
      "eval_samples_per_second": 7.409,
      "eval_steps_per_second": 0.412,
      "step": 2800
    },
    {
      "epoch": 0.8165585598717294,
      "grad_norm": 28.10056883840461,
      "learning_rate": 8.610983795867983e-07,
      "loss": 0.6877,
      "step": 2801
    },
    {
      "epoch": 0.8168500838131332,
      "grad_norm": 22.51128769167331,
      "learning_rate": 8.584520901884608e-07,
      "loss": 0.5617,
      "step": 2802
    },
    {
      "epoch": 0.8171416077545368,
      "grad_norm": 19.37687202844531,
      "learning_rate": 8.558094913798448e-07,
      "loss": 0.609,
      "step": 2803
    },
    {
      "epoch": 0.8174331316959406,
      "grad_norm": 25.68332329133898,
      "learning_rate": 8.531705855158024e-07,
      "loss": 0.7774,
      "step": 2804
    },
    {
      "epoch": 0.8177246556373442,
      "grad_norm": 62.95330228235775,
      "learning_rate": 8.505353749479001e-07,
      "loss": 0.725,
      "step": 2805
    },
    {
      "epoch": 0.818016179578748,
      "grad_norm": 27.142808850617286,
      "learning_rate": 8.479038620244089e-07,
      "loss": 0.7088,
      "step": 2806
    },
    {
      "epoch": 0.8183077035201516,
      "grad_norm": 59.42752093428065,
      "learning_rate": 8.452760490903073e-07,
      "loss": 0.7465,
      "step": 2807
    },
    {
      "epoch": 0.8185992274615553,
      "grad_norm": 22.620418525374394,
      "learning_rate": 8.426519384872733e-07,
      "loss": 0.5663,
      "step": 2808
    },
    {
      "epoch": 0.818890751402959,
      "grad_norm": 31.436634028728086,
      "learning_rate": 8.400315325536907e-07,
      "loss": 0.6292,
      "step": 2809
    },
    {
      "epoch": 0.8191822753443626,
      "grad_norm": 26.31738693924747,
      "learning_rate": 8.374148336246352e-07,
      "loss": 0.8082,
      "step": 2810
    },
    {
      "epoch": 0.8194737992857664,
      "grad_norm": 26.50068828058768,
      "learning_rate": 8.348018440318845e-07,
      "loss": 0.7442,
      "step": 2811
    },
    {
      "epoch": 0.81976532322717,
      "grad_norm": 44.43924108779055,
      "learning_rate": 8.321925661039088e-07,
      "loss": 0.5965,
      "step": 2812
    },
    {
      "epoch": 0.8200568471685737,
      "grad_norm": 22.65772862422224,
      "learning_rate": 8.295870021658714e-07,
      "loss": 0.6284,
      "step": 2813
    },
    {
      "epoch": 0.8203483711099774,
      "grad_norm": 35.968381684961486,
      "learning_rate": 8.269851545396279e-07,
      "loss": 0.6847,
      "step": 2814
    },
    {
      "epoch": 0.8206398950513811,
      "grad_norm": 43.839842259716626,
      "learning_rate": 8.243870255437147e-07,
      "loss": 0.6617,
      "step": 2815
    },
    {
      "epoch": 0.8209314189927848,
      "grad_norm": 55.040618705851585,
      "learning_rate": 8.217926174933665e-07,
      "loss": 0.7455,
      "step": 2816
    },
    {
      "epoch": 0.8212229429341885,
      "grad_norm": 22.75529243622935,
      "learning_rate": 8.192019327004908e-07,
      "loss": 0.7229,
      "step": 2817
    },
    {
      "epoch": 0.8215144668755922,
      "grad_norm": 32.59224332681238,
      "learning_rate": 8.166149734736845e-07,
      "loss": 0.7006,
      "step": 2818
    },
    {
      "epoch": 0.8218059908169958,
      "grad_norm": 31.81039783428369,
      "learning_rate": 8.140317421182203e-07,
      "loss": 0.7013,
      "step": 2819
    },
    {
      "epoch": 0.8220975147583995,
      "grad_norm": 23.190282652888385,
      "learning_rate": 8.114522409360531e-07,
      "loss": 0.6839,
      "step": 2820
    },
    {
      "epoch": 0.8223890386998032,
      "grad_norm": 25.943216305631143,
      "learning_rate": 8.088764722258097e-07,
      "loss": 0.6414,
      "step": 2821
    },
    {
      "epoch": 0.8226805626412069,
      "grad_norm": 19.42645761857317,
      "learning_rate": 8.063044382827945e-07,
      "loss": 0.7463,
      "step": 2822
    },
    {
      "epoch": 0.8229720865826106,
      "grad_norm": 43.22254753722598,
      "learning_rate": 8.037361413989797e-07,
      "loss": 0.7408,
      "step": 2823
    },
    {
      "epoch": 0.8232636105240143,
      "grad_norm": 24.611360854159766,
      "learning_rate": 8.011715838630107e-07,
      "loss": 0.6832,
      "step": 2824
    },
    {
      "epoch": 0.823555134465418,
      "grad_norm": 23.545700900560664,
      "learning_rate": 7.986107679602001e-07,
      "loss": 0.8163,
      "step": 2825
    },
    {
      "epoch": 0.8238466584068217,
      "grad_norm": 49.01941255545092,
      "learning_rate": 7.960536959725252e-07,
      "loss": 0.6563,
      "step": 2826
    },
    {
      "epoch": 0.8241381823482253,
      "grad_norm": 23.45638515104786,
      "learning_rate": 7.935003701786287e-07,
      "loss": 0.7655,
      "step": 2827
    },
    {
      "epoch": 0.8244297062896291,
      "grad_norm": 22.999842441732547,
      "learning_rate": 7.909507928538107e-07,
      "loss": 0.4856,
      "step": 2828
    },
    {
      "epoch": 0.8247212302310327,
      "grad_norm": 20.70103524463604,
      "learning_rate": 7.884049662700388e-07,
      "loss": 0.7047,
      "step": 2829
    },
    {
      "epoch": 0.8250127541724364,
      "grad_norm": 22.20883511012978,
      "learning_rate": 7.858628926959311e-07,
      "loss": 0.573,
      "step": 2830
    },
    {
      "epoch": 0.8253042781138401,
      "grad_norm": 146.48736304358616,
      "learning_rate": 7.833245743967638e-07,
      "loss": 0.6858,
      "step": 2831
    },
    {
      "epoch": 0.8255958020552437,
      "grad_norm": 18.421944745997564,
      "learning_rate": 7.807900136344676e-07,
      "loss": 0.5699,
      "step": 2832
    },
    {
      "epoch": 0.8258873259966475,
      "grad_norm": 32.54832789167716,
      "learning_rate": 7.782592126676242e-07,
      "loss": 0.6545,
      "step": 2833
    },
    {
      "epoch": 0.8261788499380511,
      "grad_norm": 28.22693264451793,
      "learning_rate": 7.757321737514645e-07,
      "loss": 0.8735,
      "step": 2834
    },
    {
      "epoch": 0.8264703738794549,
      "grad_norm": 34.4678456207575,
      "learning_rate": 7.732088991378689e-07,
      "loss": 0.7562,
      "step": 2835
    },
    {
      "epoch": 0.8267618978208585,
      "grad_norm": 23.677750726067078,
      "learning_rate": 7.706893910753571e-07,
      "loss": 0.781,
      "step": 2836
    },
    {
      "epoch": 0.8270534217622623,
      "grad_norm": 26.108784500596816,
      "learning_rate": 7.681736518091032e-07,
      "loss": 0.6493,
      "step": 2837
    },
    {
      "epoch": 0.8273449457036659,
      "grad_norm": 19.94401206693248,
      "learning_rate": 7.656616835809122e-07,
      "loss": 0.6541,
      "step": 2838
    },
    {
      "epoch": 0.8276364696450697,
      "grad_norm": 46.629413654860514,
      "learning_rate": 7.631534886292353e-07,
      "loss": 0.6691,
      "step": 2839
    },
    {
      "epoch": 0.8279279935864733,
      "grad_norm": 342.2459154702359,
      "learning_rate": 7.606490691891577e-07,
      "loss": 0.7313,
      "step": 2840
    },
    {
      "epoch": 0.8282195175278769,
      "grad_norm": 21.921378112973898,
      "learning_rate": 7.581484274924028e-07,
      "loss": 0.656,
      "step": 2841
    },
    {
      "epoch": 0.8285110414692807,
      "grad_norm": 34.05915691987391,
      "learning_rate": 7.556515657673274e-07,
      "loss": 0.872,
      "step": 2842
    },
    {
      "epoch": 0.8288025654106843,
      "grad_norm": 27.696199764862936,
      "learning_rate": 7.531584862389162e-07,
      "loss": 0.7075,
      "step": 2843
    },
    {
      "epoch": 0.8290940893520881,
      "grad_norm": 27.799182398636702,
      "learning_rate": 7.506691911287883e-07,
      "loss": 0.6503,
      "step": 2844
    },
    {
      "epoch": 0.8293856132934917,
      "grad_norm": 35.98284463451275,
      "learning_rate": 7.481836826551875e-07,
      "loss": 0.776,
      "step": 2845
    },
    {
      "epoch": 0.8296771372348954,
      "grad_norm": 24.181671731825087,
      "learning_rate": 7.457019630329848e-07,
      "loss": 0.8186,
      "step": 2846
    },
    {
      "epoch": 0.8299686611762991,
      "grad_norm": 32.758306810468966,
      "learning_rate": 7.43224034473674e-07,
      "loss": 0.6468,
      "step": 2847
    },
    {
      "epoch": 0.8302601851177028,
      "grad_norm": 21.406684288431546,
      "learning_rate": 7.407498991853729e-07,
      "loss": 0.7383,
      "step": 2848
    },
    {
      "epoch": 0.8305517090591065,
      "grad_norm": 23.00446287141877,
      "learning_rate": 7.382795593728131e-07,
      "loss": 0.6908,
      "step": 2849
    },
    {
      "epoch": 0.8308432330005102,
      "grad_norm": 32.18420791546144,
      "learning_rate": 7.358130172373523e-07,
      "loss": 0.7931,
      "step": 2850
    },
    {
      "epoch": 0.8308432330005102,
      "eval_loss": 1.11776602268219,
      "eval_runtime": 2415.2034,
      "eval_samples_per_second": 7.151,
      "eval_steps_per_second": 0.397,
      "step": 2850
    },
    {
      "epoch": 0.8311347569419139,
      "grad_norm": 24.372431870579028,
      "learning_rate": 7.333502749769561e-07,
      "loss": 0.6176,
      "step": 2851
    },
    {
      "epoch": 0.8314262808833175,
      "grad_norm": 35.957065649219004,
      "learning_rate": 7.308913347862112e-07,
      "loss": 0.6719,
      "step": 2852
    },
    {
      "epoch": 0.8317178048247212,
      "grad_norm": 27.01966149566141,
      "learning_rate": 7.284361988563099e-07,
      "loss": 0.8237,
      "step": 2853
    },
    {
      "epoch": 0.8320093287661249,
      "grad_norm": 25.363691746214673,
      "learning_rate": 7.259848693750582e-07,
      "loss": 0.8022,
      "step": 2854
    },
    {
      "epoch": 0.8323008527075286,
      "grad_norm": 25.618130645610186,
      "learning_rate": 7.235373485268693e-07,
      "loss": 0.7349,
      "step": 2855
    },
    {
      "epoch": 0.8325923766489323,
      "grad_norm": 37.7493363364309,
      "learning_rate": 7.210936384927631e-07,
      "loss": 0.6676,
      "step": 2856
    },
    {
      "epoch": 0.832883900590336,
      "grad_norm": 33.05672678777025,
      "learning_rate": 7.186537414503614e-07,
      "loss": 0.6632,
      "step": 2857
    },
    {
      "epoch": 0.8331754245317397,
      "grad_norm": 22.500036914071433,
      "learning_rate": 7.162176595738895e-07,
      "loss": 0.6095,
      "step": 2858
    },
    {
      "epoch": 0.8334669484731434,
      "grad_norm": 22.845713771062588,
      "learning_rate": 7.137853950341744e-07,
      "loss": 0.8483,
      "step": 2859
    },
    {
      "epoch": 0.833758472414547,
      "grad_norm": 23.998991752789248,
      "learning_rate": 7.113569499986401e-07,
      "loss": 0.8882,
      "step": 2860
    },
    {
      "epoch": 0.8340499963559508,
      "grad_norm": 31.00280390809461,
      "learning_rate": 7.089323266313069e-07,
      "loss": 0.7584,
      "step": 2861
    },
    {
      "epoch": 0.8343415202973544,
      "grad_norm": 23.93778842276359,
      "learning_rate": 7.065115270927875e-07,
      "loss": 0.7308,
      "step": 2862
    },
    {
      "epoch": 0.8346330442387581,
      "grad_norm": 26.036712958686316,
      "learning_rate": 7.040945535402932e-07,
      "loss": 0.7482,
      "step": 2863
    },
    {
      "epoch": 0.8349245681801618,
      "grad_norm": 17.836543950721097,
      "learning_rate": 7.01681408127618e-07,
      "loss": 0.6253,
      "step": 2864
    },
    {
      "epoch": 0.8352160921215654,
      "grad_norm": 26.541017082576207,
      "learning_rate": 6.992720930051527e-07,
      "loss": 0.7035,
      "step": 2865
    },
    {
      "epoch": 0.8355076160629692,
      "grad_norm": 37.71035614169283,
      "learning_rate": 6.968666103198679e-07,
      "loss": 0.839,
      "step": 2866
    },
    {
      "epoch": 0.8357991400043728,
      "grad_norm": 26.02350954698636,
      "learning_rate": 6.944649622153227e-07,
      "loss": 0.6068,
      "step": 2867
    },
    {
      "epoch": 0.8360906639457766,
      "grad_norm": 19.560850290107624,
      "learning_rate": 6.920671508316584e-07,
      "loss": 0.5963,
      "step": 2868
    },
    {
      "epoch": 0.8363821878871802,
      "grad_norm": 21.143403319674885,
      "learning_rate": 6.896731783055987e-07,
      "loss": 0.8258,
      "step": 2869
    },
    {
      "epoch": 0.836673711828584,
      "grad_norm": 42.251920264309994,
      "learning_rate": 6.872830467704417e-07,
      "loss": 0.9098,
      "step": 2870
    },
    {
      "epoch": 0.8369652357699876,
      "grad_norm": 26.83674337578194,
      "learning_rate": 6.848967583560706e-07,
      "loss": 0.6503,
      "step": 2871
    },
    {
      "epoch": 0.8372567597113914,
      "grad_norm": 40.470774665732286,
      "learning_rate": 6.825143151889358e-07,
      "loss": 0.804,
      "step": 2872
    },
    {
      "epoch": 0.837548283652795,
      "grad_norm": 18.49244522784788,
      "learning_rate": 6.801357193920666e-07,
      "loss": 0.6603,
      "step": 2873
    },
    {
      "epoch": 0.8378398075941986,
      "grad_norm": 15.858507736193589,
      "learning_rate": 6.777609730850615e-07,
      "loss": 0.5716,
      "step": 2874
    },
    {
      "epoch": 0.8381313315356024,
      "grad_norm": 45.3042270949448,
      "learning_rate": 6.7539007838409e-07,
      "loss": 0.8116,
      "step": 2875
    },
    {
      "epoch": 0.838422855477006,
      "grad_norm": 20.51210897939657,
      "learning_rate": 6.730230374018886e-07,
      "loss": 0.7089,
      "step": 2876
    },
    {
      "epoch": 0.8387143794184098,
      "grad_norm": 43.55634408475096,
      "learning_rate": 6.706598522477575e-07,
      "loss": 0.6096,
      "step": 2877
    },
    {
      "epoch": 0.8390059033598134,
      "grad_norm": 21.77171599489055,
      "learning_rate": 6.683005250275676e-07,
      "loss": 0.7542,
      "step": 2878
    },
    {
      "epoch": 0.8392974273012171,
      "grad_norm": 35.1869714080554,
      "learning_rate": 6.659450578437432e-07,
      "loss": 0.7541,
      "step": 2879
    },
    {
      "epoch": 0.8395889512426208,
      "grad_norm": 27.07135347376504,
      "learning_rate": 6.635934527952747e-07,
      "loss": 0.562,
      "step": 2880
    },
    {
      "epoch": 0.8398804751840245,
      "grad_norm": 28.88050941147928,
      "learning_rate": 6.612457119777099e-07,
      "loss": 0.7843,
      "step": 2881
    },
    {
      "epoch": 0.8401719991254282,
      "grad_norm": 17.60987457852725,
      "learning_rate": 6.589018374831529e-07,
      "loss": 0.7055,
      "step": 2882
    },
    {
      "epoch": 0.8404635230668319,
      "grad_norm": 20.777894281288987,
      "learning_rate": 6.565618314002597e-07,
      "loss": 0.5671,
      "step": 2883
    },
    {
      "epoch": 0.8407550470082356,
      "grad_norm": 32.860874650384936,
      "learning_rate": 6.542256958142456e-07,
      "loss": 0.7423,
      "step": 2884
    },
    {
      "epoch": 0.8410465709496392,
      "grad_norm": 66.13587901272159,
      "learning_rate": 6.518934328068682e-07,
      "loss": 0.6784,
      "step": 2885
    },
    {
      "epoch": 0.8413380948910429,
      "grad_norm": 20.151011761858076,
      "learning_rate": 6.495650444564433e-07,
      "loss": 0.6364,
      "step": 2886
    },
    {
      "epoch": 0.8416296188324466,
      "grad_norm": 33.55963810079468,
      "learning_rate": 6.472405328378262e-07,
      "loss": 0.7077,
      "step": 2887
    },
    {
      "epoch": 0.8419211427738503,
      "grad_norm": 32.83457320276842,
      "learning_rate": 6.449199000224221e-07,
      "loss": 0.6975,
      "step": 2888
    },
    {
      "epoch": 0.842212666715254,
      "grad_norm": 28.72241601251178,
      "learning_rate": 6.426031480781786e-07,
      "loss": 0.6826,
      "step": 2889
    },
    {
      "epoch": 0.8425041906566577,
      "grad_norm": 42.952959215269786,
      "learning_rate": 6.402902790695842e-07,
      "loss": 0.691,
      "step": 2890
    },
    {
      "epoch": 0.8427957145980614,
      "grad_norm": 28.297584810531195,
      "learning_rate": 6.379812950576697e-07,
      "loss": 0.7713,
      "step": 2891
    },
    {
      "epoch": 0.8430872385394651,
      "grad_norm": 32.38024469921455,
      "learning_rate": 6.356761980999998e-07,
      "loss": 0.6108,
      "step": 2892
    },
    {
      "epoch": 0.8433787624808687,
      "grad_norm": 84.04175985725952,
      "learning_rate": 6.333749902506791e-07,
      "loss": 0.5485,
      "step": 2893
    },
    {
      "epoch": 0.8436702864222724,
      "grad_norm": 53.553998289824364,
      "learning_rate": 6.310776735603452e-07,
      "loss": 0.6887,
      "step": 2894
    },
    {
      "epoch": 0.8439618103636761,
      "grad_norm": 25.839107446637584,
      "learning_rate": 6.287842500761688e-07,
      "loss": 0.7733,
      "step": 2895
    },
    {
      "epoch": 0.8442533343050798,
      "grad_norm": 42.417675187118945,
      "learning_rate": 6.264947218418482e-07,
      "loss": 0.6175,
      "step": 2896
    },
    {
      "epoch": 0.8445448582464835,
      "grad_norm": 21.443737723329985,
      "learning_rate": 6.242090908976167e-07,
      "loss": 0.7331,
      "step": 2897
    },
    {
      "epoch": 0.8448363821878871,
      "grad_norm": 22.362029026234097,
      "learning_rate": 6.219273592802278e-07,
      "loss": 0.7276,
      "step": 2898
    },
    {
      "epoch": 0.8451279061292909,
      "grad_norm": 52.85280252198074,
      "learning_rate": 6.196495290229676e-07,
      "loss": 0.5905,
      "step": 2899
    },
    {
      "epoch": 0.8454194300706945,
      "grad_norm": 23.591826365963982,
      "learning_rate": 6.173756021556377e-07,
      "loss": 0.7907,
      "step": 2900
    },
    {
      "epoch": 0.8454194300706945,
      "eval_loss": 1.1155520677566528,
      "eval_runtime": 2416.3785,
      "eval_samples_per_second": 7.147,
      "eval_steps_per_second": 0.397,
      "step": 2900
    },
    {
      "epoch": 0.8457109540120983,
      "grad_norm": 31.14908225342624,
      "learning_rate": 6.151055807045675e-07,
      "loss": 0.6578,
      "step": 2901
    },
    {
      "epoch": 0.8460024779535019,
      "grad_norm": 22.299034884090773,
      "learning_rate": 6.128394666926035e-07,
      "loss": 0.6345,
      "step": 2902
    },
    {
      "epoch": 0.8462940018949057,
      "grad_norm": 29.514960948468,
      "learning_rate": 6.105772621391104e-07,
      "loss": 0.6535,
      "step": 2903
    },
    {
      "epoch": 0.8465855258363093,
      "grad_norm": 34.6753472068539,
      "learning_rate": 6.083189690599712e-07,
      "loss": 0.8453,
      "step": 2904
    },
    {
      "epoch": 0.8468770497777129,
      "grad_norm": 21.876413477467864,
      "learning_rate": 6.060645894675787e-07,
      "loss": 0.6799,
      "step": 2905
    },
    {
      "epoch": 0.8471685737191167,
      "grad_norm": 32.775301185157794,
      "learning_rate": 6.038141253708429e-07,
      "loss": 0.7,
      "step": 2906
    },
    {
      "epoch": 0.8474600976605203,
      "grad_norm": 32.84872075559199,
      "learning_rate": 6.015675787751829e-07,
      "loss": 0.8024,
      "step": 2907
    },
    {
      "epoch": 0.8477516216019241,
      "grad_norm": 28.081897484118553,
      "learning_rate": 5.993249516825278e-07,
      "loss": 0.8029,
      "step": 2908
    },
    {
      "epoch": 0.8480431455433277,
      "grad_norm": 25.574531175470476,
      "learning_rate": 5.970862460913102e-07,
      "loss": 0.8076,
      "step": 2909
    },
    {
      "epoch": 0.8483346694847315,
      "grad_norm": 33.234289974244746,
      "learning_rate": 5.948514639964748e-07,
      "loss": 0.9198,
      "step": 2910
    },
    {
      "epoch": 0.8486261934261351,
      "grad_norm": 27.23511923795875,
      "learning_rate": 5.92620607389463e-07,
      "loss": 0.7875,
      "step": 2911
    },
    {
      "epoch": 0.8489177173675388,
      "grad_norm": 24.161771077692617,
      "learning_rate": 5.903936782582253e-07,
      "loss": 0.6483,
      "step": 2912
    },
    {
      "epoch": 0.8492092413089425,
      "grad_norm": 41.06463644993408,
      "learning_rate": 5.881706785872055e-07,
      "loss": 0.6873,
      "step": 2913
    },
    {
      "epoch": 0.8495007652503462,
      "grad_norm": 28.473555084222024,
      "learning_rate": 5.859516103573492e-07,
      "loss": 0.6064,
      "step": 2914
    },
    {
      "epoch": 0.8497922891917499,
      "grad_norm": 21.128986924544602,
      "learning_rate": 5.837364755460995e-07,
      "loss": 0.6562,
      "step": 2915
    },
    {
      "epoch": 0.8500838131331535,
      "grad_norm": 54.595111801671464,
      "learning_rate": 5.815252761273927e-07,
      "loss": 0.7577,
      "step": 2916
    },
    {
      "epoch": 0.8503753370745573,
      "grad_norm": 21.806073596258255,
      "learning_rate": 5.793180140716581e-07,
      "loss": 0.8084,
      "step": 2917
    },
    {
      "epoch": 0.8506668610159609,
      "grad_norm": 85.17175076115011,
      "learning_rate": 5.771146913458187e-07,
      "loss": 0.6379,
      "step": 2918
    },
    {
      "epoch": 0.8509583849573646,
      "grad_norm": 33.89094494391786,
      "learning_rate": 5.749153099132831e-07,
      "loss": 0.6822,
      "step": 2919
    },
    {
      "epoch": 0.8512499088987683,
      "grad_norm": 20.818830379958953,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.4929,
      "step": 2920
    },
    {
      "epoch": 0.851541432840172,
      "grad_norm": 40.97260728653176,
      "learning_rate": 5.705283787642074e-07,
      "loss": 0.8064,
      "step": 2921
    },
    {
      "epoch": 0.8518329567815757,
      "grad_norm": 27.731784993934184,
      "learning_rate": 5.683408329569212e-07,
      "loss": 0.6922,
      "step": 2922
    },
    {
      "epoch": 0.8521244807229794,
      "grad_norm": 47.80601328563637,
      "learning_rate": 5.661572362614454e-07,
      "loss": 0.8264,
      "step": 2923
    },
    {
      "epoch": 0.852416004664383,
      "grad_norm": 25.071056618192877,
      "learning_rate": 5.6397759062361e-07,
      "loss": 0.5882,
      "step": 2924
    },
    {
      "epoch": 0.8527075286057868,
      "grad_norm": 29.365655512288875,
      "learning_rate": 5.61801897985731e-07,
      "loss": 0.5839,
      "step": 2925
    },
    {
      "epoch": 0.8529990525471904,
      "grad_norm": 23.191569897677027,
      "learning_rate": 5.596301602865938e-07,
      "loss": 0.8754,
      "step": 2926
    },
    {
      "epoch": 0.8532905764885941,
      "grad_norm": 26.065009785780262,
      "learning_rate": 5.574623794614658e-07,
      "loss": 0.6412,
      "step": 2927
    },
    {
      "epoch": 0.8535821004299978,
      "grad_norm": 44.15187592251112,
      "learning_rate": 5.55298557442085e-07,
      "loss": 0.7325,
      "step": 2928
    },
    {
      "epoch": 0.8538736243714015,
      "grad_norm": 45.38157174559824,
      "learning_rate": 5.531386961566626e-07,
      "loss": 0.7289,
      "step": 2929
    },
    {
      "epoch": 0.8541651483128052,
      "grad_norm": 29.079500804652263,
      "learning_rate": 5.509827975298809e-07,
      "loss": 0.7055,
      "step": 2930
    },
    {
      "epoch": 0.8544566722542088,
      "grad_norm": 21.402566042175305,
      "learning_rate": 5.488308634828915e-07,
      "loss": 0.7062,
      "step": 2931
    },
    {
      "epoch": 0.8547481961956126,
      "grad_norm": 34.906481966776134,
      "learning_rate": 5.466828959333087e-07,
      "loss": 0.7206,
      "step": 2932
    },
    {
      "epoch": 0.8550397201370162,
      "grad_norm": 21.492663978141522,
      "learning_rate": 5.445388967952197e-07,
      "loss": 0.6642,
      "step": 2933
    },
    {
      "epoch": 0.85533124407842,
      "grad_norm": 22.664565488084676,
      "learning_rate": 5.423988679791686e-07,
      "loss": 0.6834,
      "step": 2934
    },
    {
      "epoch": 0.8556227680198236,
      "grad_norm": 23.237675281816703,
      "learning_rate": 5.402628113921654e-07,
      "loss": 0.8372,
      "step": 2935
    },
    {
      "epoch": 0.8559142919612274,
      "grad_norm": 22.236488461394774,
      "learning_rate": 5.381307289376786e-07,
      "loss": 0.885,
      "step": 2936
    },
    {
      "epoch": 0.856205815902631,
      "grad_norm": 26.336388042854743,
      "learning_rate": 5.360026225156362e-07,
      "loss": 0.5809,
      "step": 2937
    },
    {
      "epoch": 0.8564973398440346,
      "grad_norm": 37.03373984907522,
      "learning_rate": 5.338784940224239e-07,
      "loss": 0.6432,
      "step": 2938
    },
    {
      "epoch": 0.8567888637854384,
      "grad_norm": 34.50578416739401,
      "learning_rate": 5.317583453508795e-07,
      "loss": 0.645,
      "step": 2939
    },
    {
      "epoch": 0.857080387726842,
      "grad_norm": 23.88737880603752,
      "learning_rate": 5.296421783902972e-07,
      "loss": 0.7455,
      "step": 2940
    },
    {
      "epoch": 0.8573719116682458,
      "grad_norm": 26.592490083148192,
      "learning_rate": 5.275299950264223e-07,
      "loss": 0.6843,
      "step": 2941
    },
    {
      "epoch": 0.8576634356096494,
      "grad_norm": 36.33756558789781,
      "learning_rate": 5.254217971414499e-07,
      "loss": 0.6602,
      "step": 2942
    },
    {
      "epoch": 0.8579549595510532,
      "grad_norm": 37.381157271632446,
      "learning_rate": 5.233175866140244e-07,
      "loss": 0.6312,
      "step": 2943
    },
    {
      "epoch": 0.8582464834924568,
      "grad_norm": 25.43718470223928,
      "learning_rate": 5.212173653192365e-07,
      "loss": 0.8031,
      "step": 2944
    },
    {
      "epoch": 0.8585380074338606,
      "grad_norm": 70.8448362368915,
      "learning_rate": 5.191211351286196e-07,
      "loss": 0.6573,
      "step": 2945
    },
    {
      "epoch": 0.8588295313752642,
      "grad_norm": 35.21772145617214,
      "learning_rate": 5.170288979101573e-07,
      "loss": 0.7788,
      "step": 2946
    },
    {
      "epoch": 0.8591210553166679,
      "grad_norm": 26.52643768636276,
      "learning_rate": 5.149406555282672e-07,
      "loss": 0.6666,
      "step": 2947
    },
    {
      "epoch": 0.8594125792580716,
      "grad_norm": 31.281479638692367,
      "learning_rate": 5.128564098438116e-07,
      "loss": 0.8976,
      "step": 2948
    },
    {
      "epoch": 0.8597041031994752,
      "grad_norm": 26.062191623063704,
      "learning_rate": 5.1077616271409e-07,
      "loss": 0.6702,
      "step": 2949
    },
    {
      "epoch": 0.859995627140879,
      "grad_norm": 23.511626009555915,
      "learning_rate": 5.086999159928391e-07,
      "loss": 0.551,
      "step": 2950
    },
    {
      "epoch": 0.859995627140879,
      "eval_loss": 1.115296483039856,
      "eval_runtime": 2202.1281,
      "eval_samples_per_second": 7.842,
      "eval_steps_per_second": 0.436,
      "step": 2950
    },
    {
      "epoch": 0.8602871510822826,
      "grad_norm": 23.720565555148557,
      "learning_rate": 5.066276715302304e-07,
      "loss": 0.7359,
      "step": 2951
    },
    {
      "epoch": 0.8605786750236863,
      "grad_norm": 24.692732833063125,
      "learning_rate": 5.045594311728708e-07,
      "loss": 0.6302,
      "step": 2952
    },
    {
      "epoch": 0.86087019896509,
      "grad_norm": 34.63107251741356,
      "learning_rate": 5.024951967637948e-07,
      "loss": 0.5595,
      "step": 2953
    },
    {
      "epoch": 0.8611617229064937,
      "grad_norm": 117.24004354793709,
      "learning_rate": 5.00434970142471e-07,
      "loss": 0.7263,
      "step": 2954
    },
    {
      "epoch": 0.8614532468478974,
      "grad_norm": 25.585995886328902,
      "learning_rate": 4.983787531447959e-07,
      "loss": 0.5995,
      "step": 2955
    },
    {
      "epoch": 0.8617447707893011,
      "grad_norm": 18.60910852334587,
      "learning_rate": 4.963265476030916e-07,
      "loss": 0.5951,
      "step": 2956
    },
    {
      "epoch": 0.8620362947307048,
      "grad_norm": 38.52294518008702,
      "learning_rate": 4.942783553461072e-07,
      "loss": 0.7376,
      "step": 2957
    },
    {
      "epoch": 0.8623278186721084,
      "grad_norm": 46.679348041810165,
      "learning_rate": 4.922341781990131e-07,
      "loss": 0.6547,
      "step": 2958
    },
    {
      "epoch": 0.8626193426135121,
      "grad_norm": 47.06887674646795,
      "learning_rate": 4.901940179834059e-07,
      "loss": 0.7766,
      "step": 2959
    },
    {
      "epoch": 0.8629108665549158,
      "grad_norm": 43.08383091934358,
      "learning_rate": 4.881578765172979e-07,
      "loss": 0.6263,
      "step": 2960
    },
    {
      "epoch": 0.8632023904963195,
      "grad_norm": 23.172817573151406,
      "learning_rate": 4.861257556151238e-07,
      "loss": 0.6347,
      "step": 2961
    },
    {
      "epoch": 0.8634939144377232,
      "grad_norm": 22.713707294861745,
      "learning_rate": 4.840976570877332e-07,
      "loss": 0.5587,
      "step": 2962
    },
    {
      "epoch": 0.8637854383791269,
      "grad_norm": 21.578685736685834,
      "learning_rate": 4.820735827423928e-07,
      "loss": 0.6397,
      "step": 2963
    },
    {
      "epoch": 0.8640769623205306,
      "grad_norm": 27.357916652502244,
      "learning_rate": 4.800535343827834e-07,
      "loss": 0.6053,
      "step": 2964
    },
    {
      "epoch": 0.8643684862619343,
      "grad_norm": 37.464154280846124,
      "learning_rate": 4.780375138089976e-07,
      "loss": 0.6593,
      "step": 2965
    },
    {
      "epoch": 0.8646600102033379,
      "grad_norm": 25.257055326770836,
      "learning_rate": 4.7602552281753647e-07,
      "loss": 0.6041,
      "step": 2966
    },
    {
      "epoch": 0.8649515341447417,
      "grad_norm": 118.36062968910065,
      "learning_rate": 4.740175632013161e-07,
      "loss": 0.7033,
      "step": 2967
    },
    {
      "epoch": 0.8652430580861453,
      "grad_norm": 35.94935150642358,
      "learning_rate": 4.720136367496536e-07,
      "loss": 0.6644,
      "step": 2968
    },
    {
      "epoch": 0.865534582027549,
      "grad_norm": 27.279582539726047,
      "learning_rate": 4.700137452482767e-07,
      "loss": 0.7503,
      "step": 2969
    },
    {
      "epoch": 0.8658261059689527,
      "grad_norm": 43.786948975300334,
      "learning_rate": 4.6801789047931535e-07,
      "loss": 0.6283,
      "step": 2970
    },
    {
      "epoch": 0.8661176299103563,
      "grad_norm": 31.116487520962707,
      "learning_rate": 4.660260742213035e-07,
      "loss": 0.6479,
      "step": 2971
    },
    {
      "epoch": 0.8664091538517601,
      "grad_norm": 34.67687628888471,
      "learning_rate": 4.6403829824917643e-07,
      "loss": 0.6866,
      "step": 2972
    },
    {
      "epoch": 0.8667006777931637,
      "grad_norm": 36.485180416411076,
      "learning_rate": 4.620545643342661e-07,
      "loss": 0.5811,
      "step": 2973
    },
    {
      "epoch": 0.8669922017345675,
      "grad_norm": 38.4965971130751,
      "learning_rate": 4.6007487424430565e-07,
      "loss": 0.7712,
      "step": 2974
    },
    {
      "epoch": 0.8672837256759711,
      "grad_norm": 31.459843140846058,
      "learning_rate": 4.58099229743425e-07,
      "loss": 0.7083,
      "step": 2975
    },
    {
      "epoch": 0.8675752496173749,
      "grad_norm": 33.21051890742665,
      "learning_rate": 4.5612763259214653e-07,
      "loss": 0.5899,
      "step": 2976
    },
    {
      "epoch": 0.8678667735587785,
      "grad_norm": 28.955787795596134,
      "learning_rate": 4.5416008454738813e-07,
      "loss": 0.7097,
      "step": 2977
    },
    {
      "epoch": 0.8681582975001823,
      "grad_norm": 212.8080388727676,
      "learning_rate": 4.52196587362459e-07,
      "loss": 0.8261,
      "step": 2978
    },
    {
      "epoch": 0.8684498214415859,
      "grad_norm": 22.99653902738423,
      "learning_rate": 4.5023714278705613e-07,
      "loss": 0.5801,
      "step": 2979
    },
    {
      "epoch": 0.8687413453829895,
      "grad_norm": 44.40214591944113,
      "learning_rate": 4.4828175256727056e-07,
      "loss": 0.573,
      "step": 2980
    },
    {
      "epoch": 0.8690328693243933,
      "grad_norm": 28.79957563714463,
      "learning_rate": 4.4633041844557415e-07,
      "loss": 0.6328,
      "step": 2981
    },
    {
      "epoch": 0.8693243932657969,
      "grad_norm": 25.66717041044216,
      "learning_rate": 4.4438314216082856e-07,
      "loss": 0.5939,
      "step": 2982
    },
    {
      "epoch": 0.8696159172072007,
      "grad_norm": 19.421808688210113,
      "learning_rate": 4.424399254482781e-07,
      "loss": 0.6463,
      "step": 2983
    },
    {
      "epoch": 0.8699074411486043,
      "grad_norm": 29.021722261489106,
      "learning_rate": 4.405007700395497e-07,
      "loss": 0.796,
      "step": 2984
    },
    {
      "epoch": 0.870198965090008,
      "grad_norm": 51.8257986757839,
      "learning_rate": 4.385656776626518e-07,
      "loss": 0.4821,
      "step": 2985
    },
    {
      "epoch": 0.8704904890314117,
      "grad_norm": 29.132635293529624,
      "learning_rate": 4.3663465004196995e-07,
      "loss": 0.6016,
      "step": 2986
    },
    {
      "epoch": 0.8707820129728154,
      "grad_norm": 22.221590329833653,
      "learning_rate": 4.3470768889826985e-07,
      "loss": 0.5848,
      "step": 2987
    },
    {
      "epoch": 0.8710735369142191,
      "grad_norm": 17.695931313429373,
      "learning_rate": 4.3278479594869307e-07,
      "loss": 0.6196,
      "step": 2988
    },
    {
      "epoch": 0.8713650608556228,
      "grad_norm": 25.24543502107407,
      "learning_rate": 4.308659729067555e-07,
      "loss": 0.6703,
      "step": 2989
    },
    {
      "epoch": 0.8716565847970265,
      "grad_norm": 30.85412002759178,
      "learning_rate": 4.289512214823466e-07,
      "loss": 0.5408,
      "step": 2990
    },
    {
      "epoch": 0.8719481087384301,
      "grad_norm": 28.570047513427784,
      "learning_rate": 4.2704054338172874e-07,
      "loss": 0.7305,
      "step": 2991
    },
    {
      "epoch": 0.8722396326798338,
      "grad_norm": 35.321300255243365,
      "learning_rate": 4.251339403075294e-07,
      "loss": 0.549,
      "step": 2992
    },
    {
      "epoch": 0.8725311566212375,
      "grad_norm": 30.481775216609414,
      "learning_rate": 4.232314139587529e-07,
      "loss": 0.7034,
      "step": 2993
    },
    {
      "epoch": 0.8728226805626412,
      "grad_norm": 26.83536031706451,
      "learning_rate": 4.21332966030763e-07,
      "loss": 0.6331,
      "step": 2994
    },
    {
      "epoch": 0.8731142045040449,
      "grad_norm": 40.714192990651426,
      "learning_rate": 4.1943859821529545e-07,
      "loss": 0.5667,
      "step": 2995
    },
    {
      "epoch": 0.8734057284454486,
      "grad_norm": 27.07346285237519,
      "learning_rate": 4.175483122004448e-07,
      "loss": 0.572,
      "step": 2996
    },
    {
      "epoch": 0.8736972523868523,
      "grad_norm": 23.599181940148174,
      "learning_rate": 4.1566210967067145e-07,
      "loss": 0.6217,
      "step": 2997
    },
    {
      "epoch": 0.873988776328256,
      "grad_norm": 21.081851608959134,
      "learning_rate": 4.1377999230679646e-07,
      "loss": 0.5534,
      "step": 2998
    },
    {
      "epoch": 0.8742803002696596,
      "grad_norm": 42.44054063048175,
      "learning_rate": 4.119019617860004e-07,
      "loss": 0.8321,
      "step": 2999
    },
    {
      "epoch": 0.8745718242110634,
      "grad_norm": 37.14067270441548,
      "learning_rate": 4.100280197818207e-07,
      "loss": 0.7462,
      "step": 3000
    },
    {
      "epoch": 0.8745718242110634,
      "eval_loss": 1.1284602880477905,
      "eval_runtime": 2278.1092,
      "eval_samples_per_second": 7.581,
      "eval_steps_per_second": 0.421,
      "step": 3000
    },
    {
      "epoch": 0.874863348152467,
      "grad_norm": 32.20589777501591,
      "learning_rate": 4.081581679641539e-07,
      "loss": 0.7572,
      "step": 3001
    },
    {
      "epoch": 0.8751548720938707,
      "grad_norm": 45.060679567039486,
      "learning_rate": 4.062924079992492e-07,
      "loss": 0.7358,
      "step": 3002
    },
    {
      "epoch": 0.8754463960352744,
      "grad_norm": 24.72948991462783,
      "learning_rate": 4.0443074154971116e-07,
      "loss": 0.6045,
      "step": 3003
    },
    {
      "epoch": 0.875737919976678,
      "grad_norm": 22.303070962414438,
      "learning_rate": 4.025731702744978e-07,
      "loss": 0.6045,
      "step": 3004
    },
    {
      "epoch": 0.8760294439180818,
      "grad_norm": 34.735030458054055,
      "learning_rate": 4.007196958289122e-07,
      "loss": 0.6478,
      "step": 3005
    },
    {
      "epoch": 0.8763209678594854,
      "grad_norm": 87.3435248454574,
      "learning_rate": 3.9887031986461546e-07,
      "loss": 0.733,
      "step": 3006
    },
    {
      "epoch": 0.8766124918008892,
      "grad_norm": 30.89574707287392,
      "learning_rate": 3.9702504402960753e-07,
      "loss": 0.565,
      "step": 3007
    },
    {
      "epoch": 0.8769040157422928,
      "grad_norm": 48.99357502645376,
      "learning_rate": 3.9518386996824196e-07,
      "loss": 0.6764,
      "step": 3008
    },
    {
      "epoch": 0.8771955396836966,
      "grad_norm": 26.756000394759848,
      "learning_rate": 3.93346799321212e-07,
      "loss": 0.6856,
      "step": 3009
    },
    {
      "epoch": 0.8774870636251002,
      "grad_norm": 28.559444760312868,
      "learning_rate": 3.9151383372555696e-07,
      "loss": 0.6924,
      "step": 3010
    },
    {
      "epoch": 0.877778587566504,
      "grad_norm": 53.1946204712464,
      "learning_rate": 3.896849748146575e-07,
      "loss": 0.7497,
      "step": 3011
    },
    {
      "epoch": 0.8780701115079076,
      "grad_norm": 55.425408771740905,
      "learning_rate": 3.8786022421823497e-07,
      "loss": 0.7115,
      "step": 3012
    },
    {
      "epoch": 0.8783616354493112,
      "grad_norm": 37.10623965222852,
      "learning_rate": 3.8603958356234814e-07,
      "loss": 0.625,
      "step": 3013
    },
    {
      "epoch": 0.878653159390715,
      "grad_norm": 30.453873631222514,
      "learning_rate": 3.84223054469397e-07,
      "loss": 0.6567,
      "step": 3014
    },
    {
      "epoch": 0.8789446833321186,
      "grad_norm": 35.18442734264142,
      "learning_rate": 3.824106385581133e-07,
      "loss": 0.6542,
      "step": 3015
    },
    {
      "epoch": 0.8792362072735224,
      "grad_norm": 41.71574431328405,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.6003,
      "step": 3016
    },
    {
      "epoch": 0.879527731214926,
      "grad_norm": 67.98297761797467,
      "learning_rate": 3.787981527371576e-07,
      "loss": 0.5702,
      "step": 3017
    },
    {
      "epoch": 0.8798192551563297,
      "grad_norm": 42.19901917735898,
      "learning_rate": 3.76998086046621e-07,
      "loss": 0.7025,
      "step": 3018
    },
    {
      "epoch": 0.8801107790977334,
      "grad_norm": 23.849327532344184,
      "learning_rate": 3.752021389760213e-07,
      "loss": 0.7422,
      "step": 3019
    },
    {
      "epoch": 0.8804023030391371,
      "grad_norm": 27.939996675722305,
      "learning_rate": 3.7341031312574827e-07,
      "loss": 0.6099,
      "step": 3020
    },
    {
      "epoch": 0.8806938269805408,
      "grad_norm": 48.83346137194975,
      "learning_rate": 3.716226100925263e-07,
      "loss": 0.6405,
      "step": 3021
    },
    {
      "epoch": 0.8809853509219444,
      "grad_norm": 60.34546159455967,
      "learning_rate": 3.6983903146939894e-07,
      "loss": 0.6792,
      "step": 3022
    },
    {
      "epoch": 0.8812768748633482,
      "grad_norm": 44.646791357139826,
      "learning_rate": 3.680595788457386e-07,
      "loss": 0.5293,
      "step": 3023
    },
    {
      "epoch": 0.8815683988047518,
      "grad_norm": 53.36330381933608,
      "learning_rate": 3.6628425380723975e-07,
      "loss": 0.6603,
      "step": 3024
    },
    {
      "epoch": 0.8818599227461555,
      "grad_norm": 34.84598625054552,
      "learning_rate": 3.6451305793591963e-07,
      "loss": 0.5908,
      "step": 3025
    },
    {
      "epoch": 0.8821514466875592,
      "grad_norm": 35.65705054093808,
      "learning_rate": 3.627459928101118e-07,
      "loss": 0.7694,
      "step": 3026
    },
    {
      "epoch": 0.8824429706289629,
      "grad_norm": 56.665581085069704,
      "learning_rate": 3.609830600044756e-07,
      "loss": 0.693,
      "step": 3027
    },
    {
      "epoch": 0.8827344945703666,
      "grad_norm": 26.931727231860226,
      "learning_rate": 3.5922426108998154e-07,
      "loss": 0.7456,
      "step": 3028
    },
    {
      "epoch": 0.8830260185117703,
      "grad_norm": 33.56174677501747,
      "learning_rate": 3.5746959763392264e-07,
      "loss": 0.6005,
      "step": 3029
    },
    {
      "epoch": 0.883317542453174,
      "grad_norm": 30.18215599406507,
      "learning_rate": 3.5571907119990033e-07,
      "loss": 0.6598,
      "step": 3030
    },
    {
      "epoch": 0.8836090663945777,
      "grad_norm": 24.156748990151158,
      "learning_rate": 3.5397268334783353e-07,
      "loss": 0.6087,
      "step": 3031
    },
    {
      "epoch": 0.8839005903359813,
      "grad_norm": 39.376518365365015,
      "learning_rate": 3.522304356339529e-07,
      "loss": 0.8063,
      "step": 3032
    },
    {
      "epoch": 0.884192114277385,
      "grad_norm": 39.99777614382964,
      "learning_rate": 3.5049232961079815e-07,
      "loss": 0.6059,
      "step": 3033
    },
    {
      "epoch": 0.8844836382187887,
      "grad_norm": 28.77559323438113,
      "learning_rate": 3.4875836682722096e-07,
      "loss": 0.5424,
      "step": 3034
    },
    {
      "epoch": 0.8847751621601924,
      "grad_norm": 22.488070366568248,
      "learning_rate": 3.4702854882837644e-07,
      "loss": 0.763,
      "step": 3035
    },
    {
      "epoch": 0.8850666861015961,
      "grad_norm": 34.613939709520345,
      "learning_rate": 3.45302877155731e-07,
      "loss": 0.6535,
      "step": 3036
    },
    {
      "epoch": 0.8853582100429997,
      "grad_norm": 36.698162455449875,
      "learning_rate": 3.435813533470528e-07,
      "loss": 0.6848,
      "step": 3037
    },
    {
      "epoch": 0.8856497339844035,
      "grad_norm": 35.56326931077857,
      "learning_rate": 3.418639789364175e-07,
      "loss": 0.6222,
      "step": 3038
    },
    {
      "epoch": 0.8859412579258071,
      "grad_norm": 22.873853968192254,
      "learning_rate": 3.401507554541972e-07,
      "loss": 0.605,
      "step": 3039
    },
    {
      "epoch": 0.8862327818672109,
      "grad_norm": 35.543855654233774,
      "learning_rate": 3.3844168442707213e-07,
      "loss": 0.7841,
      "step": 3040
    },
    {
      "epoch": 0.8865243058086145,
      "grad_norm": 27.509616989872338,
      "learning_rate": 3.367367673780159e-07,
      "loss": 0.7215,
      "step": 3041
    },
    {
      "epoch": 0.8868158297500183,
      "grad_norm": 21.896504896930796,
      "learning_rate": 3.350360058263058e-07,
      "loss": 0.7689,
      "step": 3042
    },
    {
      "epoch": 0.8871073536914219,
      "grad_norm": 88.31650486955769,
      "learning_rate": 3.333394012875124e-07,
      "loss": 0.696,
      "step": 3043
    },
    {
      "epoch": 0.8873988776328255,
      "grad_norm": 36.41424067077021,
      "learning_rate": 3.3164695527350244e-07,
      "loss": 0.7388,
      "step": 3044
    },
    {
      "epoch": 0.8876904015742293,
      "grad_norm": 26.760041253473016,
      "learning_rate": 3.2995866929243893e-07,
      "loss": 0.6769,
      "step": 3045
    },
    {
      "epoch": 0.8879819255156329,
      "grad_norm": 45.75452555490048,
      "learning_rate": 3.2827454484877564e-07,
      "loss": 0.587,
      "step": 3046
    },
    {
      "epoch": 0.8882734494570367,
      "grad_norm": 22.120151081530093,
      "learning_rate": 3.2659458344325876e-07,
      "loss": 0.5872,
      "step": 3047
    },
    {
      "epoch": 0.8885649733984403,
      "grad_norm": 34.0414727077792,
      "learning_rate": 3.2491878657292643e-07,
      "loss": 0.5993,
      "step": 3048
    },
    {
      "epoch": 0.8888564973398441,
      "grad_norm": 28.876373457499312,
      "learning_rate": 3.2324715573110176e-07,
      "loss": 0.601,
      "step": 3049
    },
    {
      "epoch": 0.8891480212812477,
      "grad_norm": 18.6738696705848,
      "learning_rate": 3.215796924073983e-07,
      "loss": 0.5034,
      "step": 3050
    },
    {
      "epoch": 0.8891480212812477,
      "eval_loss": 1.1258902549743652,
      "eval_runtime": 2258.8225,
      "eval_samples_per_second": 7.646,
      "eval_steps_per_second": 0.425,
      "step": 3050
    },
    {
      "epoch": 0.8894395452226515,
      "grad_norm": 54.31990317090856,
      "learning_rate": 3.199163980877157e-07,
      "loss": 0.5985,
      "step": 3051
    },
    {
      "epoch": 0.8897310691640551,
      "grad_norm": 20.80243411803761,
      "learning_rate": 3.1825727425423837e-07,
      "loss": 0.6511,
      "step": 3052
    },
    {
      "epoch": 0.8900225931054588,
      "grad_norm": 27.036393137416745,
      "learning_rate": 3.1660232238543474e-07,
      "loss": 0.899,
      "step": 3053
    },
    {
      "epoch": 0.8903141170468625,
      "grad_norm": 24.78212174352263,
      "learning_rate": 3.149515439560524e-07,
      "loss": 0.627,
      "step": 3054
    },
    {
      "epoch": 0.8906056409882661,
      "grad_norm": 25.015057019094016,
      "learning_rate": 3.1330494043712576e-07,
      "loss": 0.7352,
      "step": 3055
    },
    {
      "epoch": 0.8908971649296699,
      "grad_norm": 32.96955464137792,
      "learning_rate": 3.116625132959633e-07,
      "loss": 0.6758,
      "step": 3056
    },
    {
      "epoch": 0.8911886888710735,
      "grad_norm": 25.794163441236826,
      "learning_rate": 3.1002426399615535e-07,
      "loss": 0.6586,
      "step": 3057
    },
    {
      "epoch": 0.8914802128124772,
      "grad_norm": 100.96368352027717,
      "learning_rate": 3.083901939975675e-07,
      "loss": 0.7225,
      "step": 3058
    },
    {
      "epoch": 0.8917717367538809,
      "grad_norm": 19.854895692786727,
      "learning_rate": 3.067603047563417e-07,
      "loss": 0.6031,
      "step": 3059
    },
    {
      "epoch": 0.8920632606952846,
      "grad_norm": 20.09354208385106,
      "learning_rate": 3.051345977248954e-07,
      "loss": 0.5386,
      "step": 3060
    },
    {
      "epoch": 0.8923547846366883,
      "grad_norm": 35.56549780673032,
      "learning_rate": 3.0351307435191824e-07,
      "loss": 0.7453,
      "step": 3061
    },
    {
      "epoch": 0.892646308578092,
      "grad_norm": 43.14984676643865,
      "learning_rate": 3.018957360823699e-07,
      "loss": 0.5213,
      "step": 3062
    },
    {
      "epoch": 0.8929378325194957,
      "grad_norm": 23.90785959328196,
      "learning_rate": 3.002825843574847e-07,
      "loss": 0.5696,
      "step": 3063
    },
    {
      "epoch": 0.8932293564608994,
      "grad_norm": 30.31626986172516,
      "learning_rate": 2.986736206147628e-07,
      "loss": 0.7507,
      "step": 3064
    },
    {
      "epoch": 0.893520880402303,
      "grad_norm": 28.42212572097968,
      "learning_rate": 2.9706884628797394e-07,
      "loss": 0.6309,
      "step": 3065
    },
    {
      "epoch": 0.8938124043437067,
      "grad_norm": 32.70795554578471,
      "learning_rate": 2.9546826280715536e-07,
      "loss": 0.692,
      "step": 3066
    },
    {
      "epoch": 0.8941039282851104,
      "grad_norm": 28.400096808890563,
      "learning_rate": 2.9387187159860664e-07,
      "loss": 0.4175,
      "step": 3067
    },
    {
      "epoch": 0.8943954522265141,
      "grad_norm": 26.89317854142067,
      "learning_rate": 2.9227967408489653e-07,
      "loss": 0.8009,
      "step": 3068
    },
    {
      "epoch": 0.8946869761679178,
      "grad_norm": 63.32011844355031,
      "learning_rate": 2.906916716848518e-07,
      "loss": 0.7049,
      "step": 3069
    },
    {
      "epoch": 0.8949785001093215,
      "grad_norm": 48.060483602500305,
      "learning_rate": 2.891078658135632e-07,
      "loss": 0.7988,
      "step": 3070
    },
    {
      "epoch": 0.8952700240507252,
      "grad_norm": 24.871836931296578,
      "learning_rate": 2.8752825788238234e-07,
      "loss": 0.6183,
      "step": 3071
    },
    {
      "epoch": 0.8955615479921288,
      "grad_norm": 60.76592406492467,
      "learning_rate": 2.859528492989194e-07,
      "loss": 0.7082,
      "step": 3072
    },
    {
      "epoch": 0.8958530719335326,
      "grad_norm": 37.36640342464754,
      "learning_rate": 2.8438164146704193e-07,
      "loss": 0.5849,
      "step": 3073
    },
    {
      "epoch": 0.8961445958749362,
      "grad_norm": 21.65235547359149,
      "learning_rate": 2.828146357868755e-07,
      "loss": 0.565,
      "step": 3074
    },
    {
      "epoch": 0.89643611981634,
      "grad_norm": 35.062905719046164,
      "learning_rate": 2.812518336547987e-07,
      "loss": 0.6898,
      "step": 3075
    },
    {
      "epoch": 0.8967276437577436,
      "grad_norm": 20.935918948313738,
      "learning_rate": 2.796932364634475e-07,
      "loss": 0.5917,
      "step": 3076
    },
    {
      "epoch": 0.8970191676991472,
      "grad_norm": 22.277836080960085,
      "learning_rate": 2.7813884560170756e-07,
      "loss": 0.7027,
      "step": 3077
    },
    {
      "epoch": 0.897310691640551,
      "grad_norm": 100.13272875082131,
      "learning_rate": 2.765886624547182e-07,
      "loss": 0.7564,
      "step": 3078
    },
    {
      "epoch": 0.8976022155819546,
      "grad_norm": 21.313235368744998,
      "learning_rate": 2.750426884038693e-07,
      "loss": 0.5056,
      "step": 3079
    },
    {
      "epoch": 0.8978937395233584,
      "grad_norm": 25.95133643878735,
      "learning_rate": 2.7350092482679836e-07,
      "loss": 0.5736,
      "step": 3080
    },
    {
      "epoch": 0.898185263464762,
      "grad_norm": 43.229704746566874,
      "learning_rate": 2.719633730973942e-07,
      "loss": 0.7102,
      "step": 3081
    },
    {
      "epoch": 0.8984767874061658,
      "grad_norm": 34.03622856592789,
      "learning_rate": 2.7043003458578685e-07,
      "loss": 0.6739,
      "step": 3082
    },
    {
      "epoch": 0.8987683113475694,
      "grad_norm": 32.65444226802558,
      "learning_rate": 2.689009106583568e-07,
      "loss": 0.7515,
      "step": 3083
    },
    {
      "epoch": 0.8990598352889732,
      "grad_norm": 79.47686724950714,
      "learning_rate": 2.673760026777272e-07,
      "loss": 0.6115,
      "step": 3084
    },
    {
      "epoch": 0.8993513592303768,
      "grad_norm": 21.031884155362032,
      "learning_rate": 2.658553120027646e-07,
      "loss": 0.6198,
      "step": 3085
    },
    {
      "epoch": 0.8996428831717804,
      "grad_norm": 24.618753858317735,
      "learning_rate": 2.6433883998857657e-07,
      "loss": 0.5793,
      "step": 3086
    },
    {
      "epoch": 0.8999344071131842,
      "grad_norm": 23.601034850955866,
      "learning_rate": 2.6282658798651274e-07,
      "loss": 0.6381,
      "step": 3087
    },
    {
      "epoch": 0.9002259310545878,
      "grad_norm": 28.95894875553642,
      "learning_rate": 2.61318557344159e-07,
      "loss": 0.6522,
      "step": 3088
    },
    {
      "epoch": 0.9005174549959916,
      "grad_norm": 26.873243134400504,
      "learning_rate": 2.59814749405346e-07,
      "loss": 0.5946,
      "step": 3089
    },
    {
      "epoch": 0.9008089789373952,
      "grad_norm": 27.09889223149305,
      "learning_rate": 2.5831516551013405e-07,
      "loss": 0.6739,
      "step": 3090
    },
    {
      "epoch": 0.901100502878799,
      "grad_norm": 33.91089302206864,
      "learning_rate": 2.5681980699482466e-07,
      "loss": 0.6839,
      "step": 3091
    },
    {
      "epoch": 0.9013920268202026,
      "grad_norm": 32.90871441902842,
      "learning_rate": 2.553286751919509e-07,
      "loss": 0.6175,
      "step": 3092
    },
    {
      "epoch": 0.9016835507616063,
      "grad_norm": 40.01894181267448,
      "learning_rate": 2.5384177143028153e-07,
      "loss": 0.6994,
      "step": 3093
    },
    {
      "epoch": 0.90197507470301,
      "grad_norm": 23.207424598731322,
      "learning_rate": 2.523590970348166e-07,
      "loss": 0.7575,
      "step": 3094
    },
    {
      "epoch": 0.9022665986444137,
      "grad_norm": 35.06463428539279,
      "learning_rate": 2.5088065332678766e-07,
      "loss": 0.5702,
      "step": 3095
    },
    {
      "epoch": 0.9025581225858174,
      "grad_norm": 33.20996863535475,
      "learning_rate": 2.4940644162365523e-07,
      "loss": 0.6399,
      "step": 3096
    },
    {
      "epoch": 0.902849646527221,
      "grad_norm": 27.96173573970252,
      "learning_rate": 2.479364632391101e-07,
      "loss": 0.6008,
      "step": 3097
    },
    {
      "epoch": 0.9031411704686247,
      "grad_norm": 23.602449381806633,
      "learning_rate": 2.46470719483069e-07,
      "loss": 0.7327,
      "step": 3098
    },
    {
      "epoch": 0.9034326944100284,
      "grad_norm": 19.049536639265206,
      "learning_rate": 2.4500921166167757e-07,
      "loss": 0.5606,
      "step": 3099
    },
    {
      "epoch": 0.9037242183514321,
      "grad_norm": 23.91533647680181,
      "learning_rate": 2.435519410773052e-07,
      "loss": 0.5927,
      "step": 3100
    },
    {
      "epoch": 0.9037242183514321,
      "eval_loss": 1.124419927597046,
      "eval_runtime": 2191.7307,
      "eval_samples_per_second": 7.88,
      "eval_steps_per_second": 0.438,
      "step": 3100
    },
    {
      "epoch": 0.9040157422928358,
      "grad_norm": 44.01605383526913,
      "learning_rate": 2.420989090285436e-07,
      "loss": 0.7131,
      "step": 3101
    },
    {
      "epoch": 0.9043072662342395,
      "grad_norm": 35.74511734739669,
      "learning_rate": 2.4065011681021266e-07,
      "loss": 0.7274,
      "step": 3102
    },
    {
      "epoch": 0.9045987901756432,
      "grad_norm": 37.53610785350808,
      "learning_rate": 2.392055657133474e-07,
      "loss": 0.5417,
      "step": 3103
    },
    {
      "epoch": 0.9048903141170469,
      "grad_norm": 38.32578955740839,
      "learning_rate": 2.3776525702520925e-07,
      "loss": 0.6868,
      "step": 3104
    },
    {
      "epoch": 0.9051818380584505,
      "grad_norm": 61.50267773067303,
      "learning_rate": 2.3632919202927595e-07,
      "loss": 0.5573,
      "step": 3105
    },
    {
      "epoch": 0.9054733619998543,
      "grad_norm": 40.73542864479468,
      "learning_rate": 2.3489737200524498e-07,
      "loss": 0.7873,
      "step": 3106
    },
    {
      "epoch": 0.9057648859412579,
      "grad_norm": 38.04168861496978,
      "learning_rate": 2.3346979822903071e-07,
      "loss": 0.5681,
      "step": 3107
    },
    {
      "epoch": 0.9060564098826616,
      "grad_norm": 49.24999005196223,
      "learning_rate": 2.3204647197276387e-07,
      "loss": 0.8132,
      "step": 3108
    },
    {
      "epoch": 0.9063479338240653,
      "grad_norm": 22.371550545719607,
      "learning_rate": 2.3062739450478822e-07,
      "loss": 0.6429,
      "step": 3109
    },
    {
      "epoch": 0.906639457765469,
      "grad_norm": 30.859905156490342,
      "learning_rate": 2.29212567089665e-07,
      "loss": 0.6795,
      "step": 3110
    },
    {
      "epoch": 0.9069309817068727,
      "grad_norm": 44.50569816544444,
      "learning_rate": 2.2780199098816513e-07,
      "loss": 0.6298,
      "step": 3111
    },
    {
      "epoch": 0.9072225056482763,
      "grad_norm": 32.29322811972597,
      "learning_rate": 2.2639566745727203e-07,
      "loss": 0.5733,
      "step": 3112
    },
    {
      "epoch": 0.9075140295896801,
      "grad_norm": 22.977257312363122,
      "learning_rate": 2.2499359775017992e-07,
      "loss": 0.6555,
      "step": 3113
    },
    {
      "epoch": 0.9078055535310837,
      "grad_norm": 20.924393286422063,
      "learning_rate": 2.2359578311629272e-07,
      "loss": 0.5128,
      "step": 3114
    },
    {
      "epoch": 0.9080970774724875,
      "grad_norm": 31.98989436076291,
      "learning_rate": 2.2220222480122232e-07,
      "loss": 0.5118,
      "step": 3115
    },
    {
      "epoch": 0.9083886014138911,
      "grad_norm": 24.117932867333625,
      "learning_rate": 2.2081292404678655e-07,
      "loss": 0.6189,
      "step": 3116
    },
    {
      "epoch": 0.9086801253552949,
      "grad_norm": 21.38801898050393,
      "learning_rate": 2.1942788209101006e-07,
      "loss": 0.6041,
      "step": 3117
    },
    {
      "epoch": 0.9089716492966985,
      "grad_norm": 53.75885047050347,
      "learning_rate": 2.1804710016812337e-07,
      "loss": 0.7616,
      "step": 3118
    },
    {
      "epoch": 0.9092631732381021,
      "grad_norm": 36.24085257338965,
      "learning_rate": 2.1667057950855996e-07,
      "loss": 0.5952,
      "step": 3119
    },
    {
      "epoch": 0.9095546971795059,
      "grad_norm": 53.83468345424161,
      "learning_rate": 2.152983213389559e-07,
      "loss": 0.5873,
      "step": 3120
    },
    {
      "epoch": 0.9098462211209095,
      "grad_norm": 25.319326920121682,
      "learning_rate": 2.1393032688214966e-07,
      "loss": 0.6367,
      "step": 3121
    },
    {
      "epoch": 0.9101377450623133,
      "grad_norm": 31.84522205522226,
      "learning_rate": 2.1256659735717777e-07,
      "loss": 0.6929,
      "step": 3122
    },
    {
      "epoch": 0.9104292690037169,
      "grad_norm": 32.844704423856,
      "learning_rate": 2.1120713397928094e-07,
      "loss": 0.7296,
      "step": 3123
    },
    {
      "epoch": 0.9107207929451206,
      "grad_norm": 33.46934656533566,
      "learning_rate": 2.0985193795989345e-07,
      "loss": 0.7661,
      "step": 3124
    },
    {
      "epoch": 0.9110123168865243,
      "grad_norm": 25.124167895406934,
      "learning_rate": 2.0850101050665038e-07,
      "loss": 0.6426,
      "step": 3125
    },
    {
      "epoch": 0.911303840827928,
      "grad_norm": 30.36262398110374,
      "learning_rate": 2.071543528233805e-07,
      "loss": 0.5925,
      "step": 3126
    },
    {
      "epoch": 0.9115953647693317,
      "grad_norm": 49.14862788389279,
      "learning_rate": 2.0581196611010933e-07,
      "loss": 0.5964,
      "step": 3127
    },
    {
      "epoch": 0.9118868887107354,
      "grad_norm": 23.63828360505573,
      "learning_rate": 2.0447385156305565e-07,
      "loss": 0.7099,
      "step": 3128
    },
    {
      "epoch": 0.9121784126521391,
      "grad_norm": 25.842852083272856,
      "learning_rate": 2.0314001037463226e-07,
      "loss": 0.7488,
      "step": 3129
    },
    {
      "epoch": 0.9124699365935427,
      "grad_norm": 51.26978147557874,
      "learning_rate": 2.0181044373344172e-07,
      "loss": 0.7283,
      "step": 3130
    },
    {
      "epoch": 0.9127614605349464,
      "grad_norm": 34.97392699946736,
      "learning_rate": 2.0048515282428017e-07,
      "loss": 0.677,
      "step": 3131
    },
    {
      "epoch": 0.9130529844763501,
      "grad_norm": 20.961571560062858,
      "learning_rate": 1.9916413882813235e-07,
      "loss": 0.6129,
      "step": 3132
    },
    {
      "epoch": 0.9133445084177538,
      "grad_norm": 26.218039668270674,
      "learning_rate": 1.9784740292217108e-07,
      "loss": 0.5673,
      "step": 3133
    },
    {
      "epoch": 0.9136360323591575,
      "grad_norm": 25.30378684695803,
      "learning_rate": 1.9653494627975888e-07,
      "loss": 0.5347,
      "step": 3134
    },
    {
      "epoch": 0.9139275563005612,
      "grad_norm": 47.83824830263459,
      "learning_rate": 1.9522677007044134e-07,
      "loss": 0.6213,
      "step": 3135
    },
    {
      "epoch": 0.9142190802419649,
      "grad_norm": 43.0834304104479,
      "learning_rate": 1.9392287545995536e-07,
      "loss": 0.7051,
      "step": 3136
    },
    {
      "epoch": 0.9145106041833686,
      "grad_norm": 35.52728390493886,
      "learning_rate": 1.9262326361021656e-07,
      "loss": 0.6532,
      "step": 3137
    },
    {
      "epoch": 0.9148021281247722,
      "grad_norm": 27.53928439401859,
      "learning_rate": 1.913279356793285e-07,
      "loss": 0.6102,
      "step": 3138
    },
    {
      "epoch": 0.915093652066176,
      "grad_norm": 44.04480895096288,
      "learning_rate": 1.90036892821574e-07,
      "loss": 0.7626,
      "step": 3139
    },
    {
      "epoch": 0.9153851760075796,
      "grad_norm": 23.03100762307096,
      "learning_rate": 1.8875013618742e-07,
      "loss": 0.666,
      "step": 3140
    },
    {
      "epoch": 0.9156766999489833,
      "grad_norm": 35.192389966410786,
      "learning_rate": 1.8746766692351259e-07,
      "loss": 0.6772,
      "step": 3141
    },
    {
      "epoch": 0.915968223890387,
      "grad_norm": 29.17089099192488,
      "learning_rate": 1.8618948617267764e-07,
      "loss": 0.6744,
      "step": 3142
    },
    {
      "epoch": 0.9162597478317906,
      "grad_norm": 40.62274432486415,
      "learning_rate": 1.8491559507391898e-07,
      "loss": 0.7551,
      "step": 3143
    },
    {
      "epoch": 0.9165512717731944,
      "grad_norm": 23.04424514149193,
      "learning_rate": 1.8364599476241862e-07,
      "loss": 0.6501,
      "step": 3144
    },
    {
      "epoch": 0.916842795714598,
      "grad_norm": 29.67031446705777,
      "learning_rate": 1.8238068636953487e-07,
      "loss": 0.6005,
      "step": 3145
    },
    {
      "epoch": 0.9171343196560018,
      "grad_norm": 42.05584836929077,
      "learning_rate": 1.8111967102280082e-07,
      "loss": 0.5802,
      "step": 3146
    },
    {
      "epoch": 0.9174258435974054,
      "grad_norm": 25.412100749531678,
      "learning_rate": 1.7986294984592535e-07,
      "loss": 0.5186,
      "step": 3147
    },
    {
      "epoch": 0.9177173675388092,
      "grad_norm": 31.783422324348216,
      "learning_rate": 1.7861052395878764e-07,
      "loss": 0.667,
      "step": 3148
    },
    {
      "epoch": 0.9180088914802128,
      "grad_norm": 47.99694923884682,
      "learning_rate": 1.773623944774444e-07,
      "loss": 0.6316,
      "step": 3149
    },
    {
      "epoch": 0.9183004154216164,
      "grad_norm": 42.565156204330336,
      "learning_rate": 1.7611856251411818e-07,
      "loss": 0.6606,
      "step": 3150
    },
    {
      "epoch": 0.9183004154216164,
      "eval_loss": 1.1253750324249268,
      "eval_runtime": 2363.4722,
      "eval_samples_per_second": 7.307,
      "eval_steps_per_second": 0.406,
      "step": 3150
    },
    {
      "epoch": 0.9185919393630202,
      "grad_norm": 144.3747381920347,
      "learning_rate": 1.7487902917720677e-07,
      "loss": 0.7197,
      "step": 3151
    },
    {
      "epoch": 0.9188834633044238,
      "grad_norm": 21.652979843592856,
      "learning_rate": 1.7364379557127387e-07,
      "loss": 0.5389,
      "step": 3152
    },
    {
      "epoch": 0.9191749872458276,
      "grad_norm": 32.49023383168415,
      "learning_rate": 1.7241286279705281e-07,
      "loss": 0.7899,
      "step": 3153
    },
    {
      "epoch": 0.9194665111872312,
      "grad_norm": 33.224951907466306,
      "learning_rate": 1.711862319514457e-07,
      "loss": 0.6796,
      "step": 3154
    },
    {
      "epoch": 0.919758035128635,
      "grad_norm": 37.800668791701625,
      "learning_rate": 1.6996390412751916e-07,
      "loss": 0.8117,
      "step": 3155
    },
    {
      "epoch": 0.9200495590700386,
      "grad_norm": 34.47531351578573,
      "learning_rate": 1.6874588041450535e-07,
      "loss": 0.6654,
      "step": 3156
    },
    {
      "epoch": 0.9203410830114424,
      "grad_norm": 49.84501695112892,
      "learning_rate": 1.6753216189780375e-07,
      "loss": 0.6947,
      "step": 3157
    },
    {
      "epoch": 0.920632606952846,
      "grad_norm": 34.11157873807131,
      "learning_rate": 1.6632274965897365e-07,
      "loss": 0.6747,
      "step": 3158
    },
    {
      "epoch": 0.9209241308942497,
      "grad_norm": 43.74539738629749,
      "learning_rate": 1.6511764477573966e-07,
      "loss": 0.7227,
      "step": 3159
    },
    {
      "epoch": 0.9212156548356534,
      "grad_norm": 53.99199803111386,
      "learning_rate": 1.639168483219872e-07,
      "loss": 0.7118,
      "step": 3160
    },
    {
      "epoch": 0.921507178777057,
      "grad_norm": 29.199340495496422,
      "learning_rate": 1.6272036136776138e-07,
      "loss": 0.6503,
      "step": 3161
    },
    {
      "epoch": 0.9217987027184608,
      "grad_norm": 22.08730801897937,
      "learning_rate": 1.6152818497926993e-07,
      "loss": 0.7569,
      "step": 3162
    },
    {
      "epoch": 0.9220902266598644,
      "grad_norm": 74.55964890655156,
      "learning_rate": 1.6034032021887523e-07,
      "loss": 0.6084,
      "step": 3163
    },
    {
      "epoch": 0.9223817506012681,
      "grad_norm": 35.65517762876125,
      "learning_rate": 1.5915676814510173e-07,
      "loss": 0.6148,
      "step": 3164
    },
    {
      "epoch": 0.9226732745426718,
      "grad_norm": 42.36283337472461,
      "learning_rate": 1.5797752981262737e-07,
      "loss": 0.6635,
      "step": 3165
    },
    {
      "epoch": 0.9229647984840755,
      "grad_norm": 60.16421394064292,
      "learning_rate": 1.5680260627228772e-07,
      "loss": 0.6981,
      "step": 3166
    },
    {
      "epoch": 0.9232563224254792,
      "grad_norm": 36.995833459821874,
      "learning_rate": 1.5563199857107358e-07,
      "loss": 0.6657,
      "step": 3167
    },
    {
      "epoch": 0.9235478463668829,
      "grad_norm": 40.62733475654203,
      "learning_rate": 1.5446570775212944e-07,
      "loss": 0.7528,
      "step": 3168
    },
    {
      "epoch": 0.9238393703082866,
      "grad_norm": 21.958768281892297,
      "learning_rate": 1.5330373485475225e-07,
      "loss": 0.7291,
      "step": 3169
    },
    {
      "epoch": 0.9241308942496903,
      "grad_norm": 22.327946695097232,
      "learning_rate": 1.5214608091439265e-07,
      "loss": 0.6077,
      "step": 3170
    },
    {
      "epoch": 0.9244224181910939,
      "grad_norm": 31.501787905764534,
      "learning_rate": 1.5099274696265097e-07,
      "loss": 0.6868,
      "step": 3171
    },
    {
      "epoch": 0.9247139421324976,
      "grad_norm": 36.7968036622436,
      "learning_rate": 1.4984373402728014e-07,
      "loss": 0.776,
      "step": 3172
    },
    {
      "epoch": 0.9250054660739013,
      "grad_norm": 50.567364927616104,
      "learning_rate": 1.4869904313217997e-07,
      "loss": 0.6771,
      "step": 3173
    },
    {
      "epoch": 0.925296990015305,
      "grad_norm": 31.34261807451617,
      "learning_rate": 1.4755867529740064e-07,
      "loss": 0.7772,
      "step": 3174
    },
    {
      "epoch": 0.9255885139567087,
      "grad_norm": 29.617005886866867,
      "learning_rate": 1.4642263153913928e-07,
      "loss": 0.6171,
      "step": 3175
    },
    {
      "epoch": 0.9258800378981124,
      "grad_norm": 49.0456308275622,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.7079,
      "step": 3176
    },
    {
      "epoch": 0.9261715618395161,
      "grad_norm": 41.299964100749676,
      "learning_rate": 1.441635202976932e-07,
      "loss": 0.6592,
      "step": 3177
    },
    {
      "epoch": 0.9264630857809197,
      "grad_norm": 90.23932835104524,
      "learning_rate": 1.4304045482763263e-07,
      "loss": 0.7323,
      "step": 3178
    },
    {
      "epoch": 0.9267546097223235,
      "grad_norm": 40.93820558752628,
      "learning_rate": 1.4192171746033768e-07,
      "loss": 0.6023,
      "step": 3179
    },
    {
      "epoch": 0.9270461336637271,
      "grad_norm": 34.86861917719355,
      "learning_rate": 1.408073091927309e-07,
      "loss": 0.7611,
      "step": 3180
    },
    {
      "epoch": 0.9273376576051309,
      "grad_norm": 49.963731657474185,
      "learning_rate": 1.396972310178768e-07,
      "loss": 0.5253,
      "step": 3181
    },
    {
      "epoch": 0.9276291815465345,
      "grad_norm": 23.749612036256984,
      "learning_rate": 1.3859148392498023e-07,
      "loss": 0.5728,
      "step": 3182
    },
    {
      "epoch": 0.9279207054879381,
      "grad_norm": 37.998745625027155,
      "learning_rate": 1.3749006889938853e-07,
      "loss": 0.5993,
      "step": 3183
    },
    {
      "epoch": 0.9282122294293419,
      "grad_norm": 29.884593337553092,
      "learning_rate": 1.3639298692258606e-07,
      "loss": 0.609,
      "step": 3184
    },
    {
      "epoch": 0.9285037533707455,
      "grad_norm": 38.26753059621982,
      "learning_rate": 1.3530023897219968e-07,
      "loss": 0.6033,
      "step": 3185
    },
    {
      "epoch": 0.9287952773121493,
      "grad_norm": 46.44949145709441,
      "learning_rate": 1.342118260219899e-07,
      "loss": 0.6873,
      "step": 3186
    },
    {
      "epoch": 0.9290868012535529,
      "grad_norm": 20.42185021751552,
      "learning_rate": 1.3312774904185755e-07,
      "loss": 0.521,
      "step": 3187
    },
    {
      "epoch": 0.9293783251949567,
      "grad_norm": 26.547699822377982,
      "learning_rate": 1.320480089978382e-07,
      "loss": 0.6894,
      "step": 3188
    },
    {
      "epoch": 0.9296698491363603,
      "grad_norm": 45.17882817607669,
      "learning_rate": 1.3097260685210277e-07,
      "loss": 0.7634,
      "step": 3189
    },
    {
      "epoch": 0.929961373077764,
      "grad_norm": 38.35139773564238,
      "learning_rate": 1.2990154356295636e-07,
      "loss": 0.6371,
      "step": 3190
    },
    {
      "epoch": 0.9302528970191677,
      "grad_norm": 27.530984829663424,
      "learning_rate": 1.2883482008483939e-07,
      "loss": 0.6147,
      "step": 3191
    },
    {
      "epoch": 0.9305444209605714,
      "grad_norm": 73.77763704569811,
      "learning_rate": 1.2777243736832202e-07,
      "loss": 0.6478,
      "step": 3192
    },
    {
      "epoch": 0.9308359449019751,
      "grad_norm": 19.995930923382403,
      "learning_rate": 1.2671439636010863e-07,
      "loss": 0.6292,
      "step": 3193
    },
    {
      "epoch": 0.9311274688433787,
      "grad_norm": 52.43359779329183,
      "learning_rate": 1.2566069800303393e-07,
      "loss": 0.6788,
      "step": 3194
    },
    {
      "epoch": 0.9314189927847825,
      "grad_norm": 44.89074850178655,
      "learning_rate": 1.2461134323606295e-07,
      "loss": 0.7856,
      "step": 3195
    },
    {
      "epoch": 0.9317105167261861,
      "grad_norm": 63.832889021582076,
      "learning_rate": 1.2356633299429044e-07,
      "loss": 0.481,
      "step": 3196
    },
    {
      "epoch": 0.9320020406675898,
      "grad_norm": 29.526584053166214,
      "learning_rate": 1.2252566820893763e-07,
      "loss": 0.6363,
      "step": 3197
    },
    {
      "epoch": 0.9322935646089935,
      "grad_norm": 27.406209568375417,
      "learning_rate": 1.2148934980735772e-07,
      "loss": 0.5829,
      "step": 3198
    },
    {
      "epoch": 0.9325850885503972,
      "grad_norm": 24.270871552550243,
      "learning_rate": 1.2045737871302588e-07,
      "loss": 0.6724,
      "step": 3199
    },
    {
      "epoch": 0.9328766124918009,
      "grad_norm": 21.083119502460573,
      "learning_rate": 1.1942975584554594e-07,
      "loss": 0.5819,
      "step": 3200
    },
    {
      "epoch": 0.9328766124918009,
      "eval_loss": 1.123437523841858,
      "eval_runtime": 2363.2162,
      "eval_samples_per_second": 7.308,
      "eval_steps_per_second": 0.406,
      "step": 3200
    },
    {
      "epoch": 0.9331681364332046,
      "grad_norm": 21.934053261352954,
      "learning_rate": 1.1840648212064765e-07,
      "loss": 0.6212,
      "step": 3201
    },
    {
      "epoch": 0.9334596603746083,
      "grad_norm": 40.1927000269186,
      "learning_rate": 1.1738755845018323e-07,
      "loss": 0.6251,
      "step": 3202
    },
    {
      "epoch": 0.933751184316012,
      "grad_norm": 32.22176452860048,
      "learning_rate": 1.1637298574213029e-07,
      "loss": 0.6445,
      "step": 3203
    },
    {
      "epoch": 0.9340427082574156,
      "grad_norm": 20.829566235867937,
      "learning_rate": 1.1536276490058784e-07,
      "loss": 0.5776,
      "step": 3204
    },
    {
      "epoch": 0.9343342321988193,
      "grad_norm": 36.05358624930919,
      "learning_rate": 1.1435689682577633e-07,
      "loss": 0.7528,
      "step": 3205
    },
    {
      "epoch": 0.934625756140223,
      "grad_norm": 22.545337211638742,
      "learning_rate": 1.1335538241404099e-07,
      "loss": 0.7468,
      "step": 3206
    },
    {
      "epoch": 0.9349172800816267,
      "grad_norm": 70.91571407623863,
      "learning_rate": 1.123582225578429e-07,
      "loss": 0.863,
      "step": 3207
    },
    {
      "epoch": 0.9352088040230304,
      "grad_norm": 30.684324307673915,
      "learning_rate": 1.1136541814576574e-07,
      "loss": 0.607,
      "step": 3208
    },
    {
      "epoch": 0.935500327964434,
      "grad_norm": 26.17787633139157,
      "learning_rate": 1.1037697006251014e-07,
      "loss": 0.7056,
      "step": 3209
    },
    {
      "epoch": 0.9357918519058378,
      "grad_norm": 34.7802058583223,
      "learning_rate": 1.0939287918889652e-07,
      "loss": 0.8479,
      "step": 3210
    },
    {
      "epoch": 0.9360833758472414,
      "grad_norm": 35.686823987076124,
      "learning_rate": 1.084131464018623e-07,
      "loss": 0.6188,
      "step": 3211
    },
    {
      "epoch": 0.9363748997886452,
      "grad_norm": 25.989646694662007,
      "learning_rate": 1.0743777257445853e-07,
      "loss": 0.5762,
      "step": 3212
    },
    {
      "epoch": 0.9366664237300488,
      "grad_norm": 23.565087413707428,
      "learning_rate": 1.0646675857585608e-07,
      "loss": 0.6956,
      "step": 3213
    },
    {
      "epoch": 0.9369579476714526,
      "grad_norm": 22.586594610752105,
      "learning_rate": 1.055001052713378e-07,
      "loss": 0.6351,
      "step": 3214
    },
    {
      "epoch": 0.9372494716128562,
      "grad_norm": 29.849886971473325,
      "learning_rate": 1.045378135223013e-07,
      "loss": 0.4913,
      "step": 3215
    },
    {
      "epoch": 0.9375409955542598,
      "grad_norm": 25.56963978096483,
      "learning_rate": 1.0357988418625897e-07,
      "loss": 0.668,
      "step": 3216
    },
    {
      "epoch": 0.9378325194956636,
      "grad_norm": 22.68695759140166,
      "learning_rate": 1.026263181168341e-07,
      "loss": 0.7123,
      "step": 3217
    },
    {
      "epoch": 0.9381240434370672,
      "grad_norm": 45.587848691138284,
      "learning_rate": 1.0167711616376196e-07,
      "loss": 0.6081,
      "step": 3218
    },
    {
      "epoch": 0.938415567378471,
      "grad_norm": 32.19505209416725,
      "learning_rate": 1.0073227917289041e-07,
      "loss": 0.7234,
      "step": 3219
    },
    {
      "epoch": 0.9387070913198746,
      "grad_norm": 269.58850807442985,
      "learning_rate": 9.979180798617538e-08,
      "loss": 0.6054,
      "step": 3220
    },
    {
      "epoch": 0.9389986152612784,
      "grad_norm": 53.115309065135406,
      "learning_rate": 9.885570344168427e-08,
      "loss": 0.6511,
      "step": 3221
    },
    {
      "epoch": 0.939290139202682,
      "grad_norm": 32.17211750714796,
      "learning_rate": 9.792396637359203e-08,
      "loss": 0.6671,
      "step": 3222
    },
    {
      "epoch": 0.9395816631440858,
      "grad_norm": 27.53046279496685,
      "learning_rate": 9.699659761218339e-08,
      "loss": 0.6837,
      "step": 3223
    },
    {
      "epoch": 0.9398731870854894,
      "grad_norm": 20.77962630081506,
      "learning_rate": 9.607359798384785e-08,
      "loss": 0.5325,
      "step": 3224
    },
    {
      "epoch": 0.940164711026893,
      "grad_norm": 32.91005361366579,
      "learning_rate": 9.515496831108418e-08,
      "loss": 0.5479,
      "step": 3225
    },
    {
      "epoch": 0.9404562349682968,
      "grad_norm": 30.07953304464434,
      "learning_rate": 9.424070941249419e-08,
      "loss": 0.6883,
      "step": 3226
    },
    {
      "epoch": 0.9407477589097004,
      "grad_norm": 36.94892249097364,
      "learning_rate": 9.333082210278788e-08,
      "loss": 0.6442,
      "step": 3227
    },
    {
      "epoch": 0.9410392828511042,
      "grad_norm": 28.2275495460225,
      "learning_rate": 9.242530719277776e-08,
      "loss": 0.6951,
      "step": 3228
    },
    {
      "epoch": 0.9413308067925078,
      "grad_norm": 30.369591381034617,
      "learning_rate": 9.152416548938059e-08,
      "loss": 0.7788,
      "step": 3229
    },
    {
      "epoch": 0.9416223307339116,
      "grad_norm": 29.799352754197766,
      "learning_rate": 9.062739779561624e-08,
      "loss": 0.6875,
      "step": 3230
    },
    {
      "epoch": 0.9419138546753152,
      "grad_norm": 47.78560558667224,
      "learning_rate": 8.97350049106055e-08,
      "loss": 0.7292,
      "step": 3231
    },
    {
      "epoch": 0.9422053786167189,
      "grad_norm": 33.17329337664296,
      "learning_rate": 8.884698762957334e-08,
      "loss": 0.827,
      "step": 3232
    },
    {
      "epoch": 0.9424969025581226,
      "grad_norm": 27.39585550631818,
      "learning_rate": 8.796334674384288e-08,
      "loss": 0.7975,
      "step": 3233
    },
    {
      "epoch": 0.9427884264995263,
      "grad_norm": 61.478150582928414,
      "learning_rate": 8.708408304083927e-08,
      "loss": 0.702,
      "step": 3234
    },
    {
      "epoch": 0.94307995044093,
      "grad_norm": 40.950288961897215,
      "learning_rate": 8.620919730408573e-08,
      "loss": 0.5594,
      "step": 3235
    },
    {
      "epoch": 0.9433714743823336,
      "grad_norm": 21.44303031304419,
      "learning_rate": 8.53386903132053e-08,
      "loss": 0.5651,
      "step": 3236
    },
    {
      "epoch": 0.9436629983237373,
      "grad_norm": 27.516411228552272,
      "learning_rate": 8.447256284391858e-08,
      "loss": 0.768,
      "step": 3237
    },
    {
      "epoch": 0.943954522265141,
      "grad_norm": 42.290330845734836,
      "learning_rate": 8.361081566804318e-08,
      "loss": 0.623,
      "step": 3238
    },
    {
      "epoch": 0.9442460462065447,
      "grad_norm": 25.79449742941676,
      "learning_rate": 8.275344955349429e-08,
      "loss": 0.688,
      "step": 3239
    },
    {
      "epoch": 0.9445375701479484,
      "grad_norm": 25.517048195480093,
      "learning_rate": 8.190046526428241e-08,
      "loss": 0.4952,
      "step": 3240
    },
    {
      "epoch": 0.9448290940893521,
      "grad_norm": 16.3604096213024,
      "learning_rate": 8.105186356051287e-08,
      "loss": 0.7823,
      "step": 3241
    },
    {
      "epoch": 0.9451206180307558,
      "grad_norm": 73.18341157640643,
      "learning_rate": 8.020764519838686e-08,
      "loss": 0.7085,
      "step": 3242
    },
    {
      "epoch": 0.9454121419721595,
      "grad_norm": 22.75646577986198,
      "learning_rate": 7.936781093019874e-08,
      "loss": 0.5355,
      "step": 3243
    },
    {
      "epoch": 0.9457036659135631,
      "grad_norm": 25.654511918459033,
      "learning_rate": 7.853236150433541e-08,
      "loss": 0.5844,
      "step": 3244
    },
    {
      "epoch": 0.9459951898549669,
      "grad_norm": 26.716687222699917,
      "learning_rate": 7.770129766527911e-08,
      "loss": 0.635,
      "step": 3245
    },
    {
      "epoch": 0.9462867137963705,
      "grad_norm": 22.5057124655812,
      "learning_rate": 7.687462015360026e-08,
      "loss": 0.6091,
      "step": 3246
    },
    {
      "epoch": 0.9465782377377742,
      "grad_norm": 23.997483321459743,
      "learning_rate": 7.605232970596288e-08,
      "loss": 0.7528,
      "step": 3247
    },
    {
      "epoch": 0.9468697616791779,
      "grad_norm": 23.260486232902316,
      "learning_rate": 7.523442705512196e-08,
      "loss": 0.5627,
      "step": 3248
    },
    {
      "epoch": 0.9471612856205816,
      "grad_norm": 25.460099248400713,
      "learning_rate": 7.442091292992116e-08,
      "loss": 0.4962,
      "step": 3249
    },
    {
      "epoch": 0.9474528095619853,
      "grad_norm": 27.180188932609628,
      "learning_rate": 7.36117880552939e-08,
      "loss": 0.6386,
      "step": 3250
    },
    {
      "epoch": 0.9474528095619853,
      "eval_loss": 1.1242187023162842,
      "eval_runtime": 2316.5126,
      "eval_samples_per_second": 7.455,
      "eval_steps_per_second": 0.414,
      "step": 3250
    },
    {
      "epoch": 0.9477443335033889,
      "grad_norm": 55.22729842820926,
      "learning_rate": 7.280705315226289e-08,
      "loss": 0.5686,
      "step": 3251
    },
    {
      "epoch": 0.9480358574447927,
      "grad_norm": 32.93691449981348,
      "learning_rate": 7.200670893793727e-08,
      "loss": 0.7005,
      "step": 3252
    },
    {
      "epoch": 0.9483273813861963,
      "grad_norm": 25.373610156075376,
      "learning_rate": 7.121075612551598e-08,
      "loss": 0.657,
      "step": 3253
    },
    {
      "epoch": 0.9486189053276001,
      "grad_norm": 56.28693335504017,
      "learning_rate": 7.041919542428221e-08,
      "loss": 0.5628,
      "step": 3254
    },
    {
      "epoch": 0.9489104292690037,
      "grad_norm": 23.86384336195989,
      "learning_rate": 6.963202753960729e-08,
      "loss": 0.7467,
      "step": 3255
    },
    {
      "epoch": 0.9492019532104075,
      "grad_norm": 41.0219448310726,
      "learning_rate": 6.884925317294678e-08,
      "loss": 0.7322,
      "step": 3256
    },
    {
      "epoch": 0.9494934771518111,
      "grad_norm": 38.84449503367865,
      "learning_rate": 6.807087302184212e-08,
      "loss": 0.5296,
      "step": 3257
    },
    {
      "epoch": 0.9497850010932147,
      "grad_norm": 20.58495797118351,
      "learning_rate": 6.72968877799185e-08,
      "loss": 0.5596,
      "step": 3258
    },
    {
      "epoch": 0.9500765250346185,
      "grad_norm": 29.48001502103592,
      "learning_rate": 6.652729813688363e-08,
      "loss": 0.6555,
      "step": 3259
    },
    {
      "epoch": 0.9503680489760221,
      "grad_norm": 31.386163843264573,
      "learning_rate": 6.576210477853007e-08,
      "loss": 0.7481,
      "step": 3260
    },
    {
      "epoch": 0.9506595729174259,
      "grad_norm": 49.16484278761032,
      "learning_rate": 6.500130838673235e-08,
      "loss": 0.5662,
      "step": 3261
    },
    {
      "epoch": 0.9509510968588295,
      "grad_norm": 37.28601268266124,
      "learning_rate": 6.424490963944597e-08,
      "loss": 0.7544,
      "step": 3262
    },
    {
      "epoch": 0.9512426208002333,
      "grad_norm": 35.101649718100184,
      "learning_rate": 6.349290921070784e-08,
      "loss": 0.6129,
      "step": 3263
    },
    {
      "epoch": 0.9515341447416369,
      "grad_norm": 31.38457162914651,
      "learning_rate": 6.274530777063747e-08,
      "loss": 0.732,
      "step": 3264
    },
    {
      "epoch": 0.9518256686830406,
      "grad_norm": 28.430725002780196,
      "learning_rate": 6.200210598542978e-08,
      "loss": 0.587,
      "step": 3265
    },
    {
      "epoch": 0.9521171926244443,
      "grad_norm": 24.812130889308566,
      "learning_rate": 6.126330451736495e-08,
      "loss": 0.551,
      "step": 3266
    },
    {
      "epoch": 0.952408716565848,
      "grad_norm": 44.5915091610394,
      "learning_rate": 6.052890402479694e-08,
      "loss": 0.5344,
      "step": 3267
    },
    {
      "epoch": 0.9527002405072517,
      "grad_norm": 30.864170795824574,
      "learning_rate": 5.97989051621617e-08,
      "loss": 0.6212,
      "step": 3268
    },
    {
      "epoch": 0.9529917644486553,
      "grad_norm": 24.116436356740277,
      "learning_rate": 5.907330857996996e-08,
      "loss": 0.5684,
      "step": 3269
    },
    {
      "epoch": 0.953283288390059,
      "grad_norm": 24.67852831910451,
      "learning_rate": 5.835211492481063e-08,
      "loss": 0.6226,
      "step": 3270
    },
    {
      "epoch": 0.9535748123314627,
      "grad_norm": 40.93550596581283,
      "learning_rate": 5.763532483934964e-08,
      "loss": 0.6433,
      "step": 3271
    },
    {
      "epoch": 0.9538663362728664,
      "grad_norm": 22.66025554100127,
      "learning_rate": 5.6922938962329364e-08,
      "loss": 0.8255,
      "step": 3272
    },
    {
      "epoch": 0.9541578602142701,
      "grad_norm": 22.475393275074506,
      "learning_rate": 5.621495792856424e-08,
      "loss": 0.7248,
      "step": 3273
    },
    {
      "epoch": 0.9544493841556738,
      "grad_norm": 33.38821015383021,
      "learning_rate": 5.551138236894793e-08,
      "loss": 0.7823,
      "step": 3274
    },
    {
      "epoch": 0.9547409080970775,
      "grad_norm": 44.322651200756106,
      "learning_rate": 5.4812212910445026e-08,
      "loss": 0.6395,
      "step": 3275
    },
    {
      "epoch": 0.9550324320384812,
      "grad_norm": 35.27401864714212,
      "learning_rate": 5.411745017609493e-08,
      "loss": 0.5856,
      "step": 3276
    },
    {
      "epoch": 0.9553239559798848,
      "grad_norm": 44.58279091104918,
      "learning_rate": 5.342709478501129e-08,
      "loss": 0.6405,
      "step": 3277
    },
    {
      "epoch": 0.9556154799212886,
      "grad_norm": 60.44867883861022,
      "learning_rate": 5.274114735237812e-08,
      "loss": 0.7311,
      "step": 3278
    },
    {
      "epoch": 0.9559070038626922,
      "grad_norm": 76.78082380524748,
      "learning_rate": 5.205960848945313e-08,
      "loss": 0.7245,
      "step": 3279
    },
    {
      "epoch": 0.9561985278040959,
      "grad_norm": 31.618043789956474,
      "learning_rate": 5.138247880356384e-08,
      "loss": 0.5174,
      "step": 3280
    },
    {
      "epoch": 0.9564900517454996,
      "grad_norm": 19.69136428221885,
      "learning_rate": 5.070975889811092e-08,
      "loss": 0.6225,
      "step": 3281
    },
    {
      "epoch": 0.9567815756869033,
      "grad_norm": 31.219825671334362,
      "learning_rate": 5.004144937256372e-08,
      "loss": 0.5024,
      "step": 3282
    },
    {
      "epoch": 0.957073099628307,
      "grad_norm": 29.28736291693928,
      "learning_rate": 4.9377550822461987e-08,
      "loss": 0.6763,
      "step": 3283
    },
    {
      "epoch": 0.9573646235697106,
      "grad_norm": 41.4198756706941,
      "learning_rate": 4.8718063839414683e-08,
      "loss": 0.6958,
      "step": 3284
    },
    {
      "epoch": 0.9576561475111144,
      "grad_norm": 32.077380675392575,
      "learning_rate": 4.806298901110062e-08,
      "loss": 0.567,
      "step": 3285
    },
    {
      "epoch": 0.957947671452518,
      "grad_norm": 27.311672766954455,
      "learning_rate": 4.741232692126396e-08,
      "loss": 0.5759,
      "step": 3286
    },
    {
      "epoch": 0.9582391953939218,
      "grad_norm": 29.976999341415276,
      "learning_rate": 4.6766078149719785e-08,
      "loss": 0.668,
      "step": 3287
    },
    {
      "epoch": 0.9585307193353254,
      "grad_norm": 28.898383052853482,
      "learning_rate": 4.612424327234966e-08,
      "loss": 0.6943,
      "step": 3288
    },
    {
      "epoch": 0.958822243276729,
      "grad_norm": 25.28087797073637,
      "learning_rate": 4.548682286109995e-08,
      "loss": 0.6941,
      "step": 3289
    },
    {
      "epoch": 0.9591137672181328,
      "grad_norm": 44.93661371010034,
      "learning_rate": 4.485381748398576e-08,
      "loss": 0.654,
      "step": 3290
    },
    {
      "epoch": 0.9594052911595364,
      "grad_norm": 80.42470459060273,
      "learning_rate": 4.4225227705086415e-08,
      "loss": 0.6792,
      "step": 3291
    },
    {
      "epoch": 0.9596968151009402,
      "grad_norm": 24.833531992984717,
      "learning_rate": 4.360105408454718e-08,
      "loss": 0.8563,
      "step": 3292
    },
    {
      "epoch": 0.9599883390423438,
      "grad_norm": 19.19962338113113,
      "learning_rate": 4.298129717857647e-08,
      "loss": 0.6405,
      "step": 3293
    },
    {
      "epoch": 0.9602798629837476,
      "grad_norm": 25.81382433948525,
      "learning_rate": 4.236595753944972e-08,
      "loss": 0.6542,
      "step": 3294
    },
    {
      "epoch": 0.9605713869251512,
      "grad_norm": 29.661900614709253,
      "learning_rate": 4.175503571550388e-08,
      "loss": 0.6704,
      "step": 3295
    },
    {
      "epoch": 0.960862910866555,
      "grad_norm": 22.06437924618055,
      "learning_rate": 4.114853225113902e-08,
      "loss": 0.6374,
      "step": 3296
    },
    {
      "epoch": 0.9611544348079586,
      "grad_norm": 29.966871121384926,
      "learning_rate": 4.0546447686819476e-08,
      "loss": 0.7489,
      "step": 3297
    },
    {
      "epoch": 0.9614459587493623,
      "grad_norm": 38.537920016840125,
      "learning_rate": 3.994878255907053e-08,
      "loss": 0.8514,
      "step": 3298
    },
    {
      "epoch": 0.961737482690766,
      "grad_norm": 28.558462926707275,
      "learning_rate": 3.9355537400480036e-08,
      "loss": 0.7323,
      "step": 3299
    },
    {
      "epoch": 0.9620290066321696,
      "grad_norm": 73.88810438924772,
      "learning_rate": 3.8766712739696786e-08,
      "loss": 0.5895,
      "step": 3300
    },
    {
      "epoch": 0.9620290066321696,
      "eval_loss": 1.1251461505889893,
      "eval_runtime": 2447.9704,
      "eval_samples_per_second": 7.055,
      "eval_steps_per_second": 0.392,
      "step": 3300
    },
    {
      "epoch": 0.9623205305735734,
      "grad_norm": 25.714586920316158,
      "learning_rate": 3.8182309101429945e-08,
      "loss": 0.7454,
      "step": 3301
    },
    {
      "epoch": 0.962612054514977,
      "grad_norm": 26.358431491879774,
      "learning_rate": 3.7602327006450166e-08,
      "loss": 0.6937,
      "step": 3302
    },
    {
      "epoch": 0.9629035784563807,
      "grad_norm": 37.79286936488409,
      "learning_rate": 3.70267669715868e-08,
      "loss": 0.6916,
      "step": 3303
    },
    {
      "epoch": 0.9631951023977844,
      "grad_norm": 53.88734940594323,
      "learning_rate": 3.645562950973014e-08,
      "loss": 0.67,
      "step": 3304
    },
    {
      "epoch": 0.9634866263391881,
      "grad_norm": 22.177505696274157,
      "learning_rate": 3.5888915129827507e-08,
      "loss": 0.6745,
      "step": 3305
    },
    {
      "epoch": 0.9637781502805918,
      "grad_norm": 27.305065995889187,
      "learning_rate": 3.5326624336886604e-08,
      "loss": 0.6857,
      "step": 3306
    },
    {
      "epoch": 0.9640696742219955,
      "grad_norm": 36.898586454764846,
      "learning_rate": 3.4768757631972186e-08,
      "loss": 0.6236,
      "step": 3307
    },
    {
      "epoch": 0.9643611981633992,
      "grad_norm": 20.465703620608835,
      "learning_rate": 3.4215315512206584e-08,
      "loss": 0.5424,
      "step": 3308
    },
    {
      "epoch": 0.9646527221048029,
      "grad_norm": 21.477191554139864,
      "learning_rate": 3.366629847076974e-08,
      "loss": 0.6699,
      "step": 3309
    },
    {
      "epoch": 0.9649442460462065,
      "grad_norm": 33.37912002671446,
      "learning_rate": 3.312170699689865e-08,
      "loss": 0.6032,
      "step": 3310
    },
    {
      "epoch": 0.9652357699876102,
      "grad_norm": 33.18615416095828,
      "learning_rate": 3.2581541575885666e-08,
      "loss": 0.8162,
      "step": 3311
    },
    {
      "epoch": 0.9655272939290139,
      "grad_norm": 48.220579801436756,
      "learning_rate": 3.204580268907909e-08,
      "loss": 0.5283,
      "step": 3312
    },
    {
      "epoch": 0.9658188178704176,
      "grad_norm": 22.103252146855354,
      "learning_rate": 3.1514490813884244e-08,
      "loss": 0.9244,
      "step": 3313
    },
    {
      "epoch": 0.9661103418118213,
      "grad_norm": 37.32786404961402,
      "learning_rate": 3.0987606423759644e-08,
      "loss": 0.5193,
      "step": 3314
    },
    {
      "epoch": 0.966401865753225,
      "grad_norm": 30.679053511880305,
      "learning_rate": 3.04651499882197e-08,
      "loss": 0.6259,
      "step": 3315
    },
    {
      "epoch": 0.9666933896946287,
      "grad_norm": 39.47262870691432,
      "learning_rate": 2.9947121972832e-08,
      "loss": 0.7076,
      "step": 3316
    },
    {
      "epoch": 0.9669849136360323,
      "grad_norm": 21.087815470514684,
      "learning_rate": 2.943352283921841e-08,
      "loss": 0.5431,
      "step": 3317
    },
    {
      "epoch": 0.9672764375774361,
      "grad_norm": 33.93762588052299,
      "learning_rate": 2.8924353045054475e-08,
      "loss": 0.6867,
      "step": 3318
    },
    {
      "epoch": 0.9675679615188397,
      "grad_norm": 49.65627102894076,
      "learning_rate": 2.8419613044067817e-08,
      "loss": 0.7372,
      "step": 3319
    },
    {
      "epoch": 0.9678594854602435,
      "grad_norm": 34.54740275782625,
      "learning_rate": 2.7919303286039202e-08,
      "loss": 0.6622,
      "step": 3320
    },
    {
      "epoch": 0.9681510094016471,
      "grad_norm": 45.91493463899189,
      "learning_rate": 2.7423424216801997e-08,
      "loss": 0.5809,
      "step": 3321
    },
    {
      "epoch": 0.9684425333430507,
      "grad_norm": 72.06980725032922,
      "learning_rate": 2.693197627823996e-08,
      "loss": 0.5817,
      "step": 3322
    },
    {
      "epoch": 0.9687340572844545,
      "grad_norm": 34.505138003495226,
      "learning_rate": 2.6444959908289437e-08,
      "loss": 0.6585,
      "step": 3323
    },
    {
      "epoch": 0.9690255812258581,
      "grad_norm": 23.10969924638249,
      "learning_rate": 2.5962375540937724e-08,
      "loss": 0.537,
      "step": 3324
    },
    {
      "epoch": 0.9693171051672619,
      "grad_norm": 68.91963942807811,
      "learning_rate": 2.5484223606221937e-08,
      "loss": 0.684,
      "step": 3325
    },
    {
      "epoch": 0.9696086291086655,
      "grad_norm": 29.84565811291528,
      "learning_rate": 2.5010504530229574e-08,
      "loss": 0.6931,
      "step": 3326
    },
    {
      "epoch": 0.9699001530500693,
      "grad_norm": 22.020810717673065,
      "learning_rate": 2.4541218735098514e-08,
      "loss": 0.5487,
      "step": 3327
    },
    {
      "epoch": 0.9701916769914729,
      "grad_norm": 23.78767429863786,
      "learning_rate": 2.4076366639015914e-08,
      "loss": 0.7085,
      "step": 3328
    },
    {
      "epoch": 0.9704832009328767,
      "grad_norm": 40.27314316127173,
      "learning_rate": 2.3615948656217082e-08,
      "loss": 0.6069,
      "step": 3329
    },
    {
      "epoch": 0.9707747248742803,
      "grad_norm": 38.094007592830856,
      "learning_rate": 2.3159965196987156e-08,
      "loss": 0.7259,
      "step": 3330
    },
    {
      "epoch": 0.971066248815684,
      "grad_norm": 53.339608031746664,
      "learning_rate": 2.2708416667659438e-08,
      "loss": 0.538,
      "step": 3331
    },
    {
      "epoch": 0.9713577727570877,
      "grad_norm": 33.729821216150434,
      "learning_rate": 2.2261303470614282e-08,
      "loss": 0.7397,
      "step": 3332
    },
    {
      "epoch": 0.9716492966984913,
      "grad_norm": 28.30731705479317,
      "learning_rate": 2.1818626004281306e-08,
      "loss": 0.4988,
      "step": 3333
    },
    {
      "epoch": 0.9719408206398951,
      "grad_norm": 38.88239534955671,
      "learning_rate": 2.1380384663135523e-08,
      "loss": 0.6914,
      "step": 3334
    },
    {
      "epoch": 0.9722323445812987,
      "grad_norm": 34.22571816621545,
      "learning_rate": 2.0946579837700653e-08,
      "loss": 0.5268,
      "step": 3335
    },
    {
      "epoch": 0.9725238685227025,
      "grad_norm": 25.41699530159258,
      "learning_rate": 2.0517211914545254e-08,
      "loss": 0.5839,
      "step": 3336
    },
    {
      "epoch": 0.9728153924641061,
      "grad_norm": 49.78387713871873,
      "learning_rate": 2.0092281276285487e-08,
      "loss": 0.5781,
      "step": 3337
    },
    {
      "epoch": 0.9731069164055098,
      "grad_norm": 43.816664903230674,
      "learning_rate": 1.967178830158234e-08,
      "loss": 0.6831,
      "step": 3338
    },
    {
      "epoch": 0.9733984403469135,
      "grad_norm": 29.866444008926877,
      "learning_rate": 1.9255733365143302e-08,
      "loss": 0.5581,
      "step": 3339
    },
    {
      "epoch": 0.9736899642883172,
      "grad_norm": 40.20724371555737,
      "learning_rate": 1.8844116837719582e-08,
      "loss": 0.5872,
      "step": 3340
    },
    {
      "epoch": 0.9739814882297209,
      "grad_norm": 22.316432216111604,
      "learning_rate": 1.8436939086109993e-08,
      "loss": 0.6006,
      "step": 3341
    },
    {
      "epoch": 0.9742730121711246,
      "grad_norm": 30.954641465989344,
      "learning_rate": 1.803420047315485e-08,
      "loss": 0.6693,
      "step": 3342
    },
    {
      "epoch": 0.9745645361125282,
      "grad_norm": 27.225101089069945,
      "learning_rate": 1.7635901357740403e-08,
      "loss": 0.6803,
      "step": 3343
    },
    {
      "epoch": 0.9748560600539319,
      "grad_norm": 76.42575983378728,
      "learning_rate": 1.724204209479663e-08,
      "loss": 0.6279,
      "step": 3344
    },
    {
      "epoch": 0.9751475839953356,
      "grad_norm": 21.671175756431364,
      "learning_rate": 1.6852623035296668e-08,
      "loss": 0.6771,
      "step": 3345
    },
    {
      "epoch": 0.9754391079367393,
      "grad_norm": 18.154098895358995,
      "learning_rate": 1.646764452625682e-08,
      "loss": 0.5983,
      "step": 3346
    },
    {
      "epoch": 0.975730631878143,
      "grad_norm": 47.6455413965278,
      "learning_rate": 1.6087106910737115e-08,
      "loss": 0.6572,
      "step": 3347
    },
    {
      "epoch": 0.9760221558195467,
      "grad_norm": 48.00799081147152,
      "learning_rate": 1.5711010527839633e-08,
      "loss": 0.5305,
      "step": 3348
    },
    {
      "epoch": 0.9763136797609504,
      "grad_norm": 35.679339795592604,
      "learning_rate": 1.5339355712709615e-08,
      "loss": 0.6619,
      "step": 3349
    },
    {
      "epoch": 0.976605203702354,
      "grad_norm": 23.906384916300134,
      "learning_rate": 1.4972142796532696e-08,
      "loss": 0.7095,
      "step": 3350
    },
    {
      "epoch": 0.976605203702354,
      "eval_loss": 1.1246883869171143,
      "eval_runtime": 2428.8183,
      "eval_samples_per_second": 7.11,
      "eval_steps_per_second": 0.395,
      "step": 3350
    },
    {
      "epoch": 0.9768967276437578,
      "grad_norm": 31.243390517256195,
      "learning_rate": 1.4609372106537678e-08,
      "loss": 0.7086,
      "step": 3351
    },
    {
      "epoch": 0.9771882515851614,
      "grad_norm": 48.37582927357887,
      "learning_rate": 1.4251043965994304e-08,
      "loss": 0.7019,
      "step": 3352
    },
    {
      "epoch": 0.9774797755265651,
      "grad_norm": 71.4720545431636,
      "learning_rate": 1.3897158694214374e-08,
      "loss": 0.6797,
      "step": 3353
    },
    {
      "epoch": 0.9777712994679688,
      "grad_norm": 53.36010638783367,
      "learning_rate": 1.3547716606548967e-08,
      "loss": 0.7026,
      "step": 3354
    },
    {
      "epoch": 0.9780628234093725,
      "grad_norm": 21.355103016154604,
      "learning_rate": 1.3202718014391214e-08,
      "loss": 0.6358,
      "step": 3355
    },
    {
      "epoch": 0.9783543473507762,
      "grad_norm": 39.368262290221,
      "learning_rate": 1.2862163225174084e-08,
      "loss": 0.6702,
      "step": 3356
    },
    {
      "epoch": 0.9786458712921798,
      "grad_norm": 45.660164075930254,
      "learning_rate": 1.2526052542370382e-08,
      "loss": 0.6078,
      "step": 3357
    },
    {
      "epoch": 0.9789373952335836,
      "grad_norm": 33.87750557278983,
      "learning_rate": 1.2194386265492742e-08,
      "loss": 0.6017,
      "step": 3358
    },
    {
      "epoch": 0.9792289191749872,
      "grad_norm": 33.411124926956546,
      "learning_rate": 1.186716469009419e-08,
      "loss": 0.6886,
      "step": 3359
    },
    {
      "epoch": 0.979520443116391,
      "grad_norm": 18.79982924889121,
      "learning_rate": 1.1544388107765924e-08,
      "loss": 0.5351,
      "step": 3360
    },
    {
      "epoch": 0.9798119670577946,
      "grad_norm": 33.524182780044484,
      "learning_rate": 1.1226056806137864e-08,
      "loss": 0.7863,
      "step": 3361
    },
    {
      "epoch": 0.9801034909991984,
      "grad_norm": 30.87889250975969,
      "learning_rate": 1.0912171068880318e-08,
      "loss": 0.8036,
      "step": 3362
    },
    {
      "epoch": 0.980395014940602,
      "grad_norm": 20.35283662196795,
      "learning_rate": 1.0602731175700654e-08,
      "loss": 0.5254,
      "step": 3363
    },
    {
      "epoch": 0.9806865388820056,
      "grad_norm": 35.9168201497054,
      "learning_rate": 1.029773740234552e-08,
      "loss": 0.6248,
      "step": 3364
    },
    {
      "epoch": 0.9809780628234094,
      "grad_norm": 29.07645085324874,
      "learning_rate": 9.997190020598623e-09,
      "loss": 0.8048,
      "step": 3365
    },
    {
      "epoch": 0.981269586764813,
      "grad_norm": 20.469612426704277,
      "learning_rate": 9.701089298281285e-09,
      "loss": 0.5847,
      "step": 3366
    },
    {
      "epoch": 0.9815611107062168,
      "grad_norm": 26.192580743348525,
      "learning_rate": 9.409435499254105e-09,
      "loss": 0.5846,
      "step": 3367
    },
    {
      "epoch": 0.9818526346476204,
      "grad_norm": 445.21447709611334,
      "learning_rate": 9.12222888341252e-09,
      "loss": 0.6051,
      "step": 3368
    },
    {
      "epoch": 0.9821441585890242,
      "grad_norm": 22.172707177300065,
      "learning_rate": 8.83946970669125e-09,
      "loss": 0.6126,
      "step": 3369
    },
    {
      "epoch": 0.9824356825304278,
      "grad_norm": 23.801337073405882,
      "learning_rate": 8.561158221060406e-09,
      "loss": 0.5308,
      "step": 3370
    },
    {
      "epoch": 0.9827272064718315,
      "grad_norm": 29.384079462709426,
      "learning_rate": 8.287294674526602e-09,
      "loss": 0.6668,
      "step": 3371
    },
    {
      "epoch": 0.9830187304132352,
      "grad_norm": 37.44709835531683,
      "learning_rate": 8.017879311134624e-09,
      "loss": 0.6375,
      "step": 3372
    },
    {
      "epoch": 0.9833102543546389,
      "grad_norm": 25.54879107162031,
      "learning_rate": 7.752912370963539e-09,
      "loss": 0.6265,
      "step": 3373
    },
    {
      "epoch": 0.9836017782960426,
      "grad_norm": 45.89052400163294,
      "learning_rate": 7.492394090128364e-09,
      "loss": 0.7061,
      "step": 3374
    },
    {
      "epoch": 0.9838933022374462,
      "grad_norm": 41.43057990548339,
      "learning_rate": 7.236324700781172e-09,
      "loss": 0.5437,
      "step": 3375
    },
    {
      "epoch": 0.98418482617885,
      "grad_norm": 43.84971678712031,
      "learning_rate": 6.98470443110888e-09,
      "loss": 0.713,
      "step": 3376
    },
    {
      "epoch": 0.9844763501202536,
      "grad_norm": 21.633063348289774,
      "learning_rate": 6.737533505333793e-09,
      "loss": 0.7358,
      "step": 3377
    },
    {
      "epoch": 0.9847678740616573,
      "grad_norm": 37.373569373911636,
      "learning_rate": 6.4948121437125035e-09,
      "loss": 0.6015,
      "step": 3378
    },
    {
      "epoch": 0.985059398003061,
      "grad_norm": 32.74349498183587,
      "learning_rate": 6.25654056253866e-09,
      "loss": 0.5132,
      "step": 3379
    },
    {
      "epoch": 0.9853509219444647,
      "grad_norm": 25.63096052427703,
      "learning_rate": 6.022718974137976e-09,
      "loss": 0.6796,
      "step": 3380
    },
    {
      "epoch": 0.9856424458858684,
      "grad_norm": 27.851989127965826,
      "learning_rate": 5.793347586873221e-09,
      "loss": 0.6725,
      "step": 3381
    },
    {
      "epoch": 0.9859339698272721,
      "grad_norm": 46.51811893685827,
      "learning_rate": 5.568426605139232e-09,
      "loss": 0.8056,
      "step": 3382
    },
    {
      "epoch": 0.9862254937686757,
      "grad_norm": 27.803859940089687,
      "learning_rate": 5.3479562293673456e-09,
      "loss": 0.5512,
      "step": 3383
    },
    {
      "epoch": 0.9865170177100795,
      "grad_norm": 39.991149057212766,
      "learning_rate": 5.131936656020409e-09,
      "loss": 0.6523,
      "step": 3384
    },
    {
      "epoch": 0.9868085416514831,
      "grad_norm": 205.72304049727575,
      "learning_rate": 4.9203680775972154e-09,
      "loss": 0.5982,
      "step": 3385
    },
    {
      "epoch": 0.9871000655928868,
      "grad_norm": 29.156118065733228,
      "learning_rate": 4.713250682629733e-09,
      "loss": 0.7453,
      "step": 3386
    },
    {
      "epoch": 0.9873915895342905,
      "grad_norm": 35.052726670869674,
      "learning_rate": 4.510584655682549e-09,
      "loss": 0.7562,
      "step": 3387
    },
    {
      "epoch": 0.9876831134756942,
      "grad_norm": 29.97555211968807,
      "learning_rate": 4.312370177353975e-09,
      "loss": 0.6012,
      "step": 3388
    },
    {
      "epoch": 0.9879746374170979,
      "grad_norm": 21.3613825537591,
      "learning_rate": 4.118607424276055e-09,
      "loss": 0.7376,
      "step": 3389
    },
    {
      "epoch": 0.9882661613585015,
      "grad_norm": 30.02057547782528,
      "learning_rate": 3.929296569112895e-09,
      "loss": 0.7116,
      "step": 3390
    },
    {
      "epoch": 0.9885576852999053,
      "grad_norm": 26.58651283073424,
      "learning_rate": 3.744437780562327e-09,
      "loss": 0.5767,
      "step": 3391
    },
    {
      "epoch": 0.9888492092413089,
      "grad_norm": 30.06158680064544,
      "learning_rate": 3.5640312233548024e-09,
      "loss": 0.6677,
      "step": 3392
    },
    {
      "epoch": 0.9891407331827127,
      "grad_norm": 31.920690062058124,
      "learning_rate": 3.3880770582522814e-09,
      "loss": 0.5816,
      "step": 3393
    },
    {
      "epoch": 0.9894322571241163,
      "grad_norm": 42.12741783391313,
      "learning_rate": 3.2165754420510063e-09,
      "loss": 0.6095,
      "step": 3394
    },
    {
      "epoch": 0.9897237810655201,
      "grad_norm": 32.803656309595176,
      "learning_rate": 3.0495265275776177e-09,
      "loss": 0.7161,
      "step": 3395
    },
    {
      "epoch": 0.9900153050069237,
      "grad_norm": 27.956849687700636,
      "learning_rate": 2.886930463691928e-09,
      "loss": 0.6314,
      "step": 3396
    },
    {
      "epoch": 0.9903068289483273,
      "grad_norm": 30.8576494035932,
      "learning_rate": 2.7287873952852596e-09,
      "loss": 0.6544,
      "step": 3397
    },
    {
      "epoch": 0.9905983528897311,
      "grad_norm": 43.66622336724336,
      "learning_rate": 2.5750974632809955e-09,
      "loss": 0.6128,
      "step": 3398
    },
    {
      "epoch": 0.9908898768311347,
      "grad_norm": 25.61499074672676,
      "learning_rate": 2.4258608046345834e-09,
      "loss": 0.7575,
      "step": 3399
    },
    {
      "epoch": 0.9911814007725385,
      "grad_norm": 29.17395176173978,
      "learning_rate": 2.2810775523329775e-09,
      "loss": 0.728,
      "step": 3400
    },
    {
      "epoch": 0.9911814007725385,
      "eval_loss": 1.124919056892395,
      "eval_runtime": 2413.8811,
      "eval_samples_per_second": 7.154,
      "eval_steps_per_second": 0.398,
      "step": 3400
    },
    {
      "epoch": 0.9914729247139421,
      "grad_norm": 20.481339379384853,
      "learning_rate": 2.140747835394641e-09,
      "loss": 0.559,
      "step": 3401
    },
    {
      "epoch": 0.9917644486553459,
      "grad_norm": 17.616368849616244,
      "learning_rate": 2.0048717788684335e-09,
      "loss": 0.6317,
      "step": 3402
    },
    {
      "epoch": 0.9920559725967495,
      "grad_norm": 27.2395948889422,
      "learning_rate": 1.8734495038363886e-09,
      "loss": 0.6814,
      "step": 3403
    },
    {
      "epoch": 0.9923474965381532,
      "grad_norm": 34.12645330377789,
      "learning_rate": 1.746481127409827e-09,
      "loss": 0.6453,
      "step": 3404
    },
    {
      "epoch": 0.9926390204795569,
      "grad_norm": 38.0035393543535,
      "learning_rate": 1.623966762732687e-09,
      "loss": 0.7668,
      "step": 3405
    },
    {
      "epoch": 0.9929305444209606,
      "grad_norm": 43.288221831007434,
      "learning_rate": 1.5059065189787502e-09,
      "loss": 0.7014,
      "step": 3406
    },
    {
      "epoch": 0.9932220683623643,
      "grad_norm": 43.77899891958455,
      "learning_rate": 1.3923005013533053e-09,
      "loss": 0.5779,
      "step": 3407
    },
    {
      "epoch": 0.9935135923037679,
      "grad_norm": 41.56558049958201,
      "learning_rate": 1.2831488110920386e-09,
      "loss": 0.7038,
      "step": 3408
    },
    {
      "epoch": 0.9938051162451716,
      "grad_norm": 23.22950974451935,
      "learning_rate": 1.1784515454621447e-09,
      "loss": 0.7042,
      "step": 3409
    },
    {
      "epoch": 0.9940966401865753,
      "grad_norm": 49.74066479333157,
      "learning_rate": 1.07820879775955e-09,
      "loss": 0.6642,
      "step": 3410
    },
    {
      "epoch": 0.994388164127979,
      "grad_norm": 38.5865570320514,
      "learning_rate": 9.824206573127992e-10,
      "loss": 0.5808,
      "step": 3411
    },
    {
      "epoch": 0.9946796880693827,
      "grad_norm": 22.466047080511178,
      "learning_rate": 8.910872094802792e-10,
      "loss": 0.7536,
      "step": 3412
    },
    {
      "epoch": 0.9949712120107864,
      "grad_norm": 54.76821436651138,
      "learning_rate": 8.042085356496643e-10,
      "loss": 0.5575,
      "step": 3413
    },
    {
      "epoch": 0.9952627359521901,
      "grad_norm": 34.797036450031946,
      "learning_rate": 7.217847132401367e-10,
      "loss": 0.6325,
      "step": 3414
    },
    {
      "epoch": 0.9955542598935938,
      "grad_norm": 24.806155945181928,
      "learning_rate": 6.438158157007213e-10,
      "loss": 0.6357,
      "step": 3415
    },
    {
      "epoch": 0.9958457838349974,
      "grad_norm": 26.701848614972803,
      "learning_rate": 5.703019125102849e-10,
      "loss": 0.9233,
      "step": 3416
    },
    {
      "epoch": 0.9961373077764011,
      "grad_norm": 35.45020289556294,
      "learning_rate": 5.012430691786474e-10,
      "loss": 0.6668,
      "step": 3417
    },
    {
      "epoch": 0.9964288317178048,
      "grad_norm": 23.053940710270574,
      "learning_rate": 4.3663934724436086e-10,
      "loss": 0.6005,
      "step": 3418
    },
    {
      "epoch": 0.9967203556592085,
      "grad_norm": 24.88627946883674,
      "learning_rate": 3.7649080427748506e-10,
      "loss": 0.5404,
      "step": 3419
    },
    {
      "epoch": 0.9970118796006122,
      "grad_norm": 73.56066127974964,
      "learning_rate": 3.20797493876257e-10,
      "loss": 0.6563,
      "step": 3420
    },
    {
      "epoch": 0.9973034035420159,
      "grad_norm": 20.68642766786077,
      "learning_rate": 2.6955946567042145e-10,
      "loss": 0.5375,
      "step": 3421
    },
    {
      "epoch": 0.9975949274834196,
      "grad_norm": 98.1020041957904,
      "learning_rate": 2.227767653190105e-10,
      "loss": 0.6619,
      "step": 3422
    },
    {
      "epoch": 0.9978864514248232,
      "grad_norm": 39.84205489607668,
      "learning_rate": 1.8044943450978847e-10,
      "loss": 0.6752,
      "step": 3423
    },
    {
      "epoch": 0.998177975366227,
      "grad_norm": 32.314176195783176,
      "learning_rate": 1.4257751096202755e-10,
      "loss": 0.5956,
      "step": 3424
    },
    {
      "epoch": 0.9984694993076306,
      "grad_norm": 22.297560894691106,
      "learning_rate": 1.0916102842428722e-10,
      "loss": 0.6082,
      "step": 3425
    },
    {
      "epoch": 0.9987610232490344,
      "grad_norm": 20.302932680051573,
      "learning_rate": 8.020001667330412e-11,
      "loss": 0.7081,
      "step": 3426
    },
    {
      "epoch": 0.999052547190438,
      "grad_norm": 27.646783992059756,
      "learning_rate": 5.5694501517322695e-11,
      "loss": 0.7686,
      "step": 3427
    },
    {
      "epoch": 0.9993440711318416,
      "grad_norm": 48.53064559499887,
      "learning_rate": 3.564450479387471e-11,
      "loss": 0.7145,
      "step": 3428
    },
    {
      "epoch": 0.9996355950732454,
      "grad_norm": 24.517889850998866,
      "learning_rate": 2.0050044368669084e-11,
      "loss": 0.6524,
      "step": 3429
    },
    {
      "epoch": 0.999927119014649,
      "grad_norm": 46.63761205048293,
      "learning_rate": 8.911134139477639e-12,
      "loss": 0.6847,
      "step": 3430
    },
    {
      "epoch": 1.0,
      "grad_norm": 46.63761205048293,
      "learning_rate": 2.2277840311391018e-12,
      "loss": 0.6397,
      "step": 3431
    },
    {
      "epoch": 1.0,
      "step": 3431,
      "total_flos": 3821508285890560.0,
      "train_loss": 0.09381582161248234,
      "train_runtime": 32091.2129,
      "train_samples_per_second": 3.848,
      "train_steps_per_second": 0.107
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3431,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 20,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3821508285890560.0,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}