{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.09684688389229731,
  "eval_steps": 500,
  "global_step": 1472,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.579272003552806e-05,
      "grad_norm": 0.5296143293380737,
      "learning_rate": 2e-05,
      "loss": 2.0928,
      "step": 1
    },
    {
      "epoch": 0.00013158544007105613,
      "grad_norm": 0.531254768371582,
      "learning_rate": 4e-05,
      "loss": 2.124,
      "step": 2
    },
    {
      "epoch": 0.00019737816010658422,
      "grad_norm": 0.5184187293052673,
      "learning_rate": 6e-05,
      "loss": 2.0397,
      "step": 3
    },
    {
      "epoch": 0.00026317088014211226,
      "grad_norm": 0.5384601950645447,
      "learning_rate": 8e-05,
      "loss": 2.1003,
      "step": 4
    },
    {
      "epoch": 0.0003289636001776403,
      "grad_norm": 0.6538830399513245,
      "learning_rate": 0.0001,
      "loss": 2.1672,
      "step": 5
    },
    {
      "epoch": 0.00039475632021316844,
      "grad_norm": 0.6739334464073181,
      "learning_rate": 9.999988534860049e-05,
      "loss": 2.0353,
      "step": 6
    },
    {
      "epoch": 0.0004605490402486965,
      "grad_norm": 0.7221142649650574,
      "learning_rate": 9.999954139492774e-05,
      "loss": 1.9623,
      "step": 7
    },
    {
      "epoch": 0.0005263417602842245,
      "grad_norm": 0.5722587704658508,
      "learning_rate": 9.999896814055916e-05,
      "loss": 1.9115,
      "step": 8
    },
    {
      "epoch": 0.0005921344803197526,
      "grad_norm": 0.5293303728103638,
      "learning_rate": 9.99981655881237e-05,
      "loss": 1.9862,
      "step": 9
    },
    {
      "epoch": 0.0006579272003552806,
      "grad_norm": 0.7691564559936523,
      "learning_rate": 9.999713374130194e-05,
      "loss": 1.7427,
      "step": 10
    },
    {
      "epoch": 0.0007237199203908088,
      "grad_norm": 0.8434930443763733,
      "learning_rate": 9.999587260482597e-05,
      "loss": 1.9118,
      "step": 11
    },
    {
      "epoch": 0.0007895126404263369,
      "grad_norm": 0.9281416535377502,
      "learning_rate": 9.999438218447944e-05,
      "loss": 2.0491,
      "step": 12
    },
    {
      "epoch": 0.0008553053604618649,
      "grad_norm": 0.7365149259567261,
      "learning_rate": 9.999266248709749e-05,
      "loss": 1.8001,
      "step": 13
    },
    {
      "epoch": 0.000921098080497393,
      "grad_norm": 0.7132240533828735,
      "learning_rate": 9.999071352056675e-05,
      "loss": 1.8975,
      "step": 14
    },
    {
      "epoch": 0.000986890800532921,
      "grad_norm": 0.644156813621521,
      "learning_rate": 9.99885352938253e-05,
      "loss": 1.8091,
      "step": 15
    },
    {
      "epoch": 0.001052683520568449,
      "grad_norm": 0.6430739164352417,
      "learning_rate": 9.99861278168626e-05,
      "loss": 1.7025,
      "step": 16
    },
    {
      "epoch": 0.0011184762406039773,
      "grad_norm": 0.7189786434173584,
      "learning_rate": 9.998349110071949e-05,
      "loss": 1.7063,
      "step": 17
    },
    {
      "epoch": 0.0011842689606395053,
      "grad_norm": 0.7468182444572449,
      "learning_rate": 9.998062515748809e-05,
      "loss": 1.7283,
      "step": 18
    },
    {
      "epoch": 0.0012500616806750333,
      "grad_norm": 0.8974145650863647,
      "learning_rate": 9.997753000031175e-05,
      "loss": 1.8559,
      "step": 19
    },
    {
      "epoch": 0.0013158544007105613,
      "grad_norm": 0.8581226468086243,
      "learning_rate": 9.99742056433851e-05,
      "loss": 1.8267,
      "step": 20
    },
    {
      "epoch": 0.0013816471207460895,
      "grad_norm": 0.8268604874610901,
      "learning_rate": 9.997065210195373e-05,
      "loss": 1.8561,
      "step": 21
    },
    {
      "epoch": 0.0014474398407816175,
      "grad_norm": 0.8176312446594238,
      "learning_rate": 9.996686939231447e-05,
      "loss": 1.9261,
      "step": 22
    },
    {
      "epoch": 0.0015132325608171455,
      "grad_norm": 0.7667636275291443,
      "learning_rate": 9.9962857531815e-05,
      "loss": 1.7929,
      "step": 23
    },
    {
      "epoch": 0.0015790252808526738,
      "grad_norm": 0.7837241291999817,
      "learning_rate": 9.995861653885393e-05,
      "loss": 1.7541,
      "step": 24
    },
    {
      "epoch": 0.0016448180008882018,
      "grad_norm": 0.8536595106124878,
      "learning_rate": 9.99541464328807e-05,
      "loss": 1.6584,
      "step": 25
    },
    {
      "epoch": 0.0017106107209237298,
      "grad_norm": 0.775619626045227,
      "learning_rate": 9.994944723439546e-05,
      "loss": 1.7213,
      "step": 26
    },
    {
      "epoch": 0.0017764034409592578,
      "grad_norm": 0.8294109106063843,
      "learning_rate": 9.994451896494901e-05,
      "loss": 1.7305,
      "step": 27
    },
    {
      "epoch": 0.001842196160994786,
      "grad_norm": 0.7727370262145996,
      "learning_rate": 9.993936164714266e-05,
      "loss": 1.7374,
      "step": 28
    },
    {
      "epoch": 0.001907988881030314,
      "grad_norm": 0.7547221183776855,
      "learning_rate": 9.993397530462818e-05,
      "loss": 1.7342,
      "step": 29
    },
    {
      "epoch": 0.001973781601065842,
      "grad_norm": 0.7097113132476807,
      "learning_rate": 9.99283599621076e-05,
      "loss": 1.7741,
      "step": 30
    },
    {
      "epoch": 0.00203957432110137,
      "grad_norm": 0.8320249319076538,
      "learning_rate": 9.992251564533322e-05,
      "loss": 1.8332,
      "step": 31
    },
    {
      "epoch": 0.002105367041136898,
      "grad_norm": 0.7366943359375,
      "learning_rate": 9.99164423811074e-05,
      "loss": 1.5746,
      "step": 32
    },
    {
      "epoch": 0.002171159761172426,
      "grad_norm": 0.7390754818916321,
      "learning_rate": 9.991014019728246e-05,
      "loss": 1.6085,
      "step": 33
    },
    {
      "epoch": 0.0022369524812079545,
      "grad_norm": 0.77167809009552,
      "learning_rate": 9.99036091227606e-05,
      "loss": 1.6211,
      "step": 34
    },
    {
      "epoch": 0.0023027452012434825,
      "grad_norm": 0.7611998319625854,
      "learning_rate": 9.989684918749365e-05,
      "loss": 1.7032,
      "step": 35
    },
    {
      "epoch": 0.0023685379212790105,
      "grad_norm": 0.7919670343399048,
      "learning_rate": 9.988986042248308e-05,
      "loss": 1.5274,
      "step": 36
    },
    {
      "epoch": 0.0024343306413145385,
      "grad_norm": 0.8527458906173706,
      "learning_rate": 9.988264285977974e-05,
      "loss": 1.5267,
      "step": 37
    },
    {
      "epoch": 0.0025001233613500666,
      "grad_norm": 0.8847199082374573,
      "learning_rate": 9.987519653248378e-05,
      "loss": 1.6056,
      "step": 38
    },
    {
      "epoch": 0.0025659160813855946,
      "grad_norm": 0.7511000633239746,
      "learning_rate": 9.986752147474449e-05,
      "loss": 1.5865,
      "step": 39
    },
    {
      "epoch": 0.0026317088014211226,
      "grad_norm": 0.8162057399749756,
      "learning_rate": 9.985961772176009e-05,
      "loss": 1.7024,
      "step": 40
    },
    {
      "epoch": 0.002697501521456651,
      "grad_norm": 0.937961757183075,
      "learning_rate": 9.985148530977767e-05,
      "loss": 1.6629,
      "step": 41
    },
    {
      "epoch": 0.002763294241492179,
      "grad_norm": 0.9329769015312195,
      "learning_rate": 9.984312427609287e-05,
      "loss": 1.7787,
      "step": 42
    },
    {
      "epoch": 0.002829086961527707,
      "grad_norm": 0.9981400370597839,
      "learning_rate": 9.983453465904992e-05,
      "loss": 1.6868,
      "step": 43
    },
    {
      "epoch": 0.002894879681563235,
      "grad_norm": 0.8342626690864563,
      "learning_rate": 9.982571649804126e-05,
      "loss": 1.6639,
      "step": 44
    },
    {
      "epoch": 0.002960672401598763,
      "grad_norm": 0.9021373391151428,
      "learning_rate": 9.981666983350746e-05,
      "loss": 1.6202,
      "step": 45
    },
    {
      "epoch": 0.003026465121634291,
      "grad_norm": 0.9746200442314148,
      "learning_rate": 9.980739470693704e-05,
      "loss": 1.5423,
      "step": 46
    },
    {
      "epoch": 0.003092257841669819,
      "grad_norm": 0.9255953431129456,
      "learning_rate": 9.979789116086625e-05,
      "loss": 1.686,
      "step": 47
    },
    {
      "epoch": 0.0031580505617053475,
      "grad_norm": 1.0396703481674194,
      "learning_rate": 9.978815923887887e-05,
      "loss": 1.6445,
      "step": 48
    },
    {
      "epoch": 0.0032238432817408755,
      "grad_norm": 1.0552101135253906,
      "learning_rate": 9.977819898560605e-05,
      "loss": 1.7063,
      "step": 49
    },
    {
      "epoch": 0.0032896360017764035,
      "grad_norm": 1.3620139360427856,
      "learning_rate": 9.976801044672608e-05,
      "loss": 1.4181,
      "step": 50
    },
    {
      "epoch": 0.0033554287218119315,
      "grad_norm": 0.8541191220283508,
      "learning_rate": 9.975759366896414e-05,
      "loss": 1.7838,
      "step": 51
    },
    {
      "epoch": 0.0034212214418474596,
      "grad_norm": 0.969934344291687,
      "learning_rate": 9.974694870009219e-05,
      "loss": 1.8725,
      "step": 52
    },
    {
      "epoch": 0.0034870141618829876,
      "grad_norm": 0.8264399170875549,
      "learning_rate": 9.973607558892864e-05,
      "loss": 1.7727,
      "step": 53
    },
    {
      "epoch": 0.0035528068819185156,
      "grad_norm": 0.7832493185997009,
      "learning_rate": 9.972497438533817e-05,
      "loss": 1.8766,
      "step": 54
    },
    {
      "epoch": 0.0036185996019540436,
      "grad_norm": 0.6866966485977173,
      "learning_rate": 9.971364514023155e-05,
      "loss": 1.7002,
      "step": 55
    },
    {
      "epoch": 0.003684392321989572,
      "grad_norm": 0.5976539850234985,
      "learning_rate": 9.970208790556532e-05,
      "loss": 1.7585,
      "step": 56
    },
    {
      "epoch": 0.0037501850420251,
      "grad_norm": 0.6232089400291443,
      "learning_rate": 9.969030273434159e-05,
      "loss": 1.8284,
      "step": 57
    },
    {
      "epoch": 0.003815977762060628,
      "grad_norm": 0.6442340612411499,
      "learning_rate": 9.967828968060783e-05,
      "loss": 1.7818,
      "step": 58
    },
    {
      "epoch": 0.003881770482096156,
      "grad_norm": 0.6159253120422363,
      "learning_rate": 9.966604879945659e-05,
      "loss": 1.6145,
      "step": 59
    },
    {
      "epoch": 0.003947563202131684,
      "grad_norm": 0.6624003648757935,
      "learning_rate": 9.965358014702519e-05,
      "loss": 1.7501,
      "step": 60
    },
    {
      "epoch": 0.004013355922167212,
      "grad_norm": 0.6175729632377625,
      "learning_rate": 9.964088378049562e-05,
      "loss": 1.7301,
      "step": 61
    },
    {
      "epoch": 0.00407914864220274,
      "grad_norm": 0.622405469417572,
      "learning_rate": 9.962795975809411e-05,
      "loss": 1.647,
      "step": 62
    },
    {
      "epoch": 0.004144941362238268,
      "grad_norm": 0.6554991602897644,
      "learning_rate": 9.961480813909094e-05,
      "loss": 1.8178,
      "step": 63
    },
    {
      "epoch": 0.004210734082273796,
      "grad_norm": 0.7180414199829102,
      "learning_rate": 9.960142898380018e-05,
      "loss": 1.9175,
      "step": 64
    },
    {
      "epoch": 0.004276526802309324,
      "grad_norm": 0.6984550356864929,
      "learning_rate": 9.958782235357938e-05,
      "loss": 1.8153,
      "step": 65
    },
    {
      "epoch": 0.004342319522344852,
      "grad_norm": 0.6315418481826782,
      "learning_rate": 9.957398831082931e-05,
      "loss": 1.6787,
      "step": 66
    },
    {
      "epoch": 0.004408112242380381,
      "grad_norm": 0.6879199743270874,
      "learning_rate": 9.955992691899367e-05,
      "loss": 1.61,
      "step": 67
    },
    {
      "epoch": 0.004473904962415909,
      "grad_norm": 0.6576867699623108,
      "learning_rate": 9.954563824255878e-05,
      "loss": 1.8015,
      "step": 68
    },
    {
      "epoch": 0.004539697682451437,
      "grad_norm": 0.656387209892273,
      "learning_rate": 9.953112234705333e-05,
      "loss": 1.8581,
      "step": 69
    },
    {
      "epoch": 0.004605490402486965,
      "grad_norm": 0.6445766091346741,
      "learning_rate": 9.9516379299048e-05,
      "loss": 1.7644,
      "step": 70
    },
    {
      "epoch": 0.004671283122522493,
      "grad_norm": 0.6688461303710938,
      "learning_rate": 9.950140916615526e-05,
      "loss": 1.7838,
      "step": 71
    },
    {
      "epoch": 0.004737075842558021,
      "grad_norm": 0.7055814862251282,
      "learning_rate": 9.948621201702896e-05,
      "loss": 1.6804,
      "step": 72
    },
    {
      "epoch": 0.004802868562593549,
      "grad_norm": 0.7650384306907654,
      "learning_rate": 9.947078792136408e-05,
      "loss": 1.7436,
      "step": 73
    },
    {
      "epoch": 0.004868661282629077,
      "grad_norm": 0.7580650448799133,
      "learning_rate": 9.945513694989639e-05,
      "loss": 1.6489,
      "step": 74
    },
    {
      "epoch": 0.004934454002664605,
      "grad_norm": 0.7199274301528931,
      "learning_rate": 9.943925917440214e-05,
      "loss": 1.7196,
      "step": 75
    },
    {
      "epoch": 0.005000246722700133,
      "grad_norm": 0.7812178134918213,
      "learning_rate": 9.942315466769765e-05,
      "loss": 1.6552,
      "step": 76
    },
    {
      "epoch": 0.005066039442735661,
      "grad_norm": 0.7464516758918762,
      "learning_rate": 9.940682350363912e-05,
      "loss": 1.7772,
      "step": 77
    },
    {
      "epoch": 0.005131832162771189,
      "grad_norm": 0.7458325028419495,
      "learning_rate": 9.939026575712218e-05,
      "loss": 1.6198,
      "step": 78
    },
    {
      "epoch": 0.005197624882806717,
      "grad_norm": 0.7330995202064514,
      "learning_rate": 9.937348150408159e-05,
      "loss": 1.7143,
      "step": 79
    },
    {
      "epoch": 0.005263417602842245,
      "grad_norm": 0.7014390826225281,
      "learning_rate": 9.935647082149086e-05,
      "loss": 1.5854,
      "step": 80
    },
    {
      "epoch": 0.005329210322877773,
      "grad_norm": 0.7711056470870972,
      "learning_rate": 9.933923378736194e-05,
      "loss": 1.8168,
      "step": 81
    },
    {
      "epoch": 0.005395003042913302,
      "grad_norm": 0.7262523174285889,
      "learning_rate": 9.932177048074483e-05,
      "loss": 1.7339,
      "step": 82
    },
    {
      "epoch": 0.00546079576294883,
      "grad_norm": 0.7787086367607117,
      "learning_rate": 9.930408098172725e-05,
      "loss": 1.6516,
      "step": 83
    },
    {
      "epoch": 0.005526588482984358,
      "grad_norm": 0.6976877450942993,
      "learning_rate": 9.928616537143421e-05,
      "loss": 1.6226,
      "step": 84
    },
    {
      "epoch": 0.005592381203019886,
      "grad_norm": 0.7678042650222778,
      "learning_rate": 9.926802373202768e-05,
      "loss": 1.6561,
      "step": 85
    },
    {
      "epoch": 0.005658173923055414,
      "grad_norm": 0.766779363155365,
      "learning_rate": 9.924965614670629e-05,
      "loss": 1.6439,
      "step": 86
    },
    {
      "epoch": 0.005723966643090942,
      "grad_norm": 0.7825470566749573,
      "learning_rate": 9.923106269970478e-05,
      "loss": 1.6673,
      "step": 87
    },
    {
      "epoch": 0.00578975936312647,
      "grad_norm": 0.8223551511764526,
      "learning_rate": 9.921224347629374e-05,
      "loss": 1.548,
      "step": 88
    },
    {
      "epoch": 0.005855552083161998,
      "grad_norm": 0.8139538168907166,
      "learning_rate": 9.91931985627792e-05,
      "loss": 1.5381,
      "step": 89
    },
    {
      "epoch": 0.005921344803197526,
      "grad_norm": 0.7926769256591797,
      "learning_rate": 9.917392804650218e-05,
      "loss": 1.6728,
      "step": 90
    },
    {
      "epoch": 0.005987137523233054,
      "grad_norm": 0.9060941934585571,
      "learning_rate": 9.915443201583834e-05,
      "loss": 1.6211,
      "step": 91
    },
    {
      "epoch": 0.006052930243268582,
      "grad_norm": 0.8837563991546631,
      "learning_rate": 9.91347105601976e-05,
      "loss": 1.5703,
      "step": 92
    },
    {
      "epoch": 0.00611872296330411,
      "grad_norm": 0.8119835257530212,
      "learning_rate": 9.911476377002364e-05,
      "loss": 1.5577,
      "step": 93
    },
    {
      "epoch": 0.006184515683339638,
      "grad_norm": 0.8880691528320312,
      "learning_rate": 9.909459173679354e-05,
      "loss": 1.7592,
      "step": 94
    },
    {
      "epoch": 0.006250308403375166,
      "grad_norm": 0.8486512303352356,
      "learning_rate": 9.907419455301741e-05,
      "loss": 1.6063,
      "step": 95
    },
    {
      "epoch": 0.006316101123410695,
      "grad_norm": 0.9571446180343628,
      "learning_rate": 9.905357231223785e-05,
      "loss": 1.711,
      "step": 96
    },
    {
      "epoch": 0.006381893843446223,
      "grad_norm": 0.9212421178817749,
      "learning_rate": 9.90327251090296e-05,
      "loss": 1.4925,
      "step": 97
    },
    {
      "epoch": 0.006447686563481751,
      "grad_norm": 0.9694377183914185,
      "learning_rate": 9.901165303899916e-05,
      "loss": 1.7433,
      "step": 98
    },
    {
      "epoch": 0.006513479283517279,
      "grad_norm": 1.1317200660705566,
      "learning_rate": 9.899035619878414e-05,
      "loss": 1.6807,
      "step": 99
    },
    {
      "epoch": 0.006579272003552807,
      "grad_norm": 1.1567927598953247,
      "learning_rate": 9.896883468605311e-05,
      "loss": 1.2842,
      "step": 100
    },
    {
      "epoch": 0.006645064723588335,
      "grad_norm": 0.6888508796691895,
      "learning_rate": 9.89470885995049e-05,
      "loss": 1.8472,
      "step": 101
    },
    {
      "epoch": 0.006710857443623863,
      "grad_norm": 0.6463817954063416,
      "learning_rate": 9.892511803886828e-05,
      "loss": 1.8032,
      "step": 102
    },
    {
      "epoch": 0.006776650163659391,
      "grad_norm": 0.6593803763389587,
      "learning_rate": 9.890292310490147e-05,
      "loss": 1.8945,
      "step": 103
    },
    {
      "epoch": 0.006842442883694919,
      "grad_norm": 0.6805739402770996,
      "learning_rate": 9.888050389939172e-05,
      "loss": 1.6301,
      "step": 104
    },
    {
      "epoch": 0.006908235603730447,
      "grad_norm": 0.6324853301048279,
      "learning_rate": 9.88578605251547e-05,
      "loss": 1.9058,
      "step": 105
    },
    {
      "epoch": 0.006974028323765975,
      "grad_norm": 0.5630888342857361,
      "learning_rate": 9.883499308603422e-05,
      "loss": 1.7052,
      "step": 106
    },
    {
      "epoch": 0.007039821043801503,
      "grad_norm": 0.5972461104393005,
      "learning_rate": 9.881190168690164e-05,
      "loss": 1.7202,
      "step": 107
    },
    {
      "epoch": 0.007105613763837031,
      "grad_norm": 0.6968611478805542,
      "learning_rate": 9.878858643365541e-05,
      "loss": 1.829,
      "step": 108
    },
    {
      "epoch": 0.007171406483872559,
      "grad_norm": 0.5739532113075256,
      "learning_rate": 9.876504743322057e-05,
      "loss": 1.7746,
      "step": 109
    },
    {
      "epoch": 0.007237199203908087,
      "grad_norm": 0.6316085457801819,
      "learning_rate": 9.874128479354832e-05,
      "loss": 1.7302,
      "step": 110
    },
    {
      "epoch": 0.007302991923943616,
      "grad_norm": 0.5930875539779663,
      "learning_rate": 9.871729862361543e-05,
      "loss": 1.625,
      "step": 111
    },
    {
      "epoch": 0.007368784643979144,
      "grad_norm": 0.6493741869926453,
      "learning_rate": 9.869308903342383e-05,
      "loss": 1.5649,
      "step": 112
    },
    {
      "epoch": 0.007434577364014672,
      "grad_norm": 0.6738541722297668,
      "learning_rate": 9.866865613400008e-05,
      "loss": 1.8187,
      "step": 113
    },
    {
      "epoch": 0.0075003700840502,
      "grad_norm": 0.6308318972587585,
      "learning_rate": 9.864400003739476e-05,
      "loss": 1.6503,
      "step": 114
    },
    {
      "epoch": 0.007566162804085728,
      "grad_norm": 0.6296387910842896,
      "learning_rate": 9.861912085668216e-05,
      "loss": 1.6963,
      "step": 115
    },
    {
      "epoch": 0.007631955524121256,
      "grad_norm": 0.6289705634117126,
      "learning_rate": 9.859401870595959e-05,
      "loss": 1.4958,
      "step": 116
    },
    {
      "epoch": 0.007697748244156784,
      "grad_norm": 0.7387570142745972,
      "learning_rate": 9.856869370034692e-05,
      "loss": 1.7227,
      "step": 117
    },
    {
      "epoch": 0.007763540964192312,
      "grad_norm": 0.6664671301841736,
      "learning_rate": 9.854314595598602e-05,
      "loss": 1.7069,
      "step": 118
    },
    {
      "epoch": 0.00782933368422784,
      "grad_norm": 0.6984542608261108,
      "learning_rate": 9.85173755900403e-05,
      "loss": 1.8907,
      "step": 119
    },
    {
      "epoch": 0.007895126404263368,
      "grad_norm": 0.6584218144416809,
      "learning_rate": 9.84913827206941e-05,
      "loss": 1.5777,
      "step": 120
    },
    {
      "epoch": 0.007960919124298896,
      "grad_norm": 0.6700417399406433,
      "learning_rate": 9.846516746715217e-05,
      "loss": 1.5087,
      "step": 121
    },
    {
      "epoch": 0.008026711844334424,
      "grad_norm": 0.7032206654548645,
      "learning_rate": 9.843872994963911e-05,
      "loss": 1.591,
      "step": 122
    },
    {
      "epoch": 0.008092504564369952,
      "grad_norm": 0.6893160939216614,
      "learning_rate": 9.841207028939889e-05,
      "loss": 1.5799,
      "step": 123
    },
    {
      "epoch": 0.00815829728440548,
      "grad_norm": 0.7209369540214539,
      "learning_rate": 9.838518860869416e-05,
      "loss": 1.5601,
      "step": 124
    },
    {
      "epoch": 0.008224090004441008,
      "grad_norm": 0.7673851251602173,
      "learning_rate": 9.835808503080585e-05,
      "loss": 1.5634,
      "step": 125
    },
    {
      "epoch": 0.008289882724476536,
      "grad_norm": 0.810266375541687,
      "learning_rate": 9.833075968003249e-05,
      "loss": 1.8813,
      "step": 126
    },
    {
      "epoch": 0.008355675444512064,
      "grad_norm": 0.6878501176834106,
      "learning_rate": 9.830321268168965e-05,
      "loss": 1.5009,
      "step": 127
    },
    {
      "epoch": 0.008421468164547592,
      "grad_norm": 0.7561343908309937,
      "learning_rate": 9.827544416210941e-05,
      "loss": 1.7062,
      "step": 128
    },
    {
      "epoch": 0.00848726088458312,
      "grad_norm": 0.6917654275894165,
      "learning_rate": 9.824745424863973e-05,
      "loss": 1.5736,
      "step": 129
    },
    {
      "epoch": 0.008553053604618648,
      "grad_norm": 0.826567530632019,
      "learning_rate": 9.821924306964398e-05,
      "loss": 1.7886,
      "step": 130
    },
    {
      "epoch": 0.008618846324654176,
      "grad_norm": 0.7179293036460876,
      "learning_rate": 9.819081075450014e-05,
      "loss": 1.6986,
      "step": 131
    },
    {
      "epoch": 0.008684639044689704,
      "grad_norm": 0.7051801085472107,
      "learning_rate": 9.816215743360044e-05,
      "loss": 1.4127,
      "step": 132
    },
    {
      "epoch": 0.008750431764725234,
      "grad_norm": 0.7013904452323914,
      "learning_rate": 9.813328323835061e-05,
      "loss": 1.7484,
      "step": 133
    },
    {
      "epoch": 0.008816224484760762,
      "grad_norm": 0.7020507454872131,
      "learning_rate": 9.810418830116932e-05,
      "loss": 1.7741,
      "step": 134
    },
    {
      "epoch": 0.00888201720479629,
      "grad_norm": 0.7591568231582642,
      "learning_rate": 9.807487275548757e-05,
      "loss": 1.5618,
      "step": 135
    },
    {
      "epoch": 0.008947809924831818,
      "grad_norm": 0.739377498626709,
      "learning_rate": 9.804533673574812e-05,
      "loss": 1.4941,
      "step": 136
    },
    {
      "epoch": 0.009013602644867346,
      "grad_norm": 0.7584704756736755,
      "learning_rate": 9.801558037740478e-05,
      "loss": 1.7772,
      "step": 137
    },
    {
      "epoch": 0.009079395364902874,
      "grad_norm": 0.7530946135520935,
      "learning_rate": 9.79856038169219e-05,
      "loss": 1.5457,
      "step": 138
    },
    {
      "epoch": 0.009145188084938402,
      "grad_norm": 0.7504359483718872,
      "learning_rate": 9.795540719177365e-05,
      "loss": 1.6163,
      "step": 139
    },
    {
      "epoch": 0.00921098080497393,
      "grad_norm": 0.8109403252601624,
      "learning_rate": 9.792499064044342e-05,
      "loss": 1.7455,
      "step": 140
    },
    {
      "epoch": 0.009276773525009458,
      "grad_norm": 0.8034817576408386,
      "learning_rate": 9.789435430242329e-05,
      "loss": 1.5284,
      "step": 141
    },
    {
      "epoch": 0.009342566245044986,
      "grad_norm": 0.8187747597694397,
      "learning_rate": 9.786349831821314e-05,
      "loss": 1.6745,
      "step": 142
    },
    {
      "epoch": 0.009408358965080514,
      "grad_norm": 0.7957119345664978,
      "learning_rate": 9.783242282932028e-05,
      "loss": 1.5745,
      "step": 143
    },
    {
      "epoch": 0.009474151685116042,
      "grad_norm": 0.8206056952476501,
      "learning_rate": 9.780112797825865e-05,
      "loss": 1.6354,
      "step": 144
    },
    {
      "epoch": 0.00953994440515157,
      "grad_norm": 0.8320123553276062,
      "learning_rate": 9.776961390854818e-05,
      "loss": 1.5993,
      "step": 145
    },
    {
      "epoch": 0.009605737125187098,
      "grad_norm": 0.8511732220649719,
      "learning_rate": 9.773788076471414e-05,
      "loss": 1.5566,
      "step": 146
    },
    {
      "epoch": 0.009671529845222626,
      "grad_norm": 0.8763437867164612,
      "learning_rate": 9.770592869228653e-05,
      "loss": 1.7509,
      "step": 147
    },
    {
      "epoch": 0.009737322565258154,
      "grad_norm": 0.9537453651428223,
      "learning_rate": 9.767375783779932e-05,
      "loss": 1.58,
      "step": 148
    },
    {
      "epoch": 0.009803115285293682,
      "grad_norm": 1.1339179277420044,
      "learning_rate": 9.764136834878986e-05,
      "loss": 1.5152,
      "step": 149
    },
    {
      "epoch": 0.00986890800532921,
      "grad_norm": 1.4353673458099365,
      "learning_rate": 9.760876037379816e-05,
      "loss": 1.4473,
      "step": 150
    },
    {
      "epoch": 0.009934700725364738,
      "grad_norm": 0.62269127368927,
      "learning_rate": 9.757593406236623e-05,
      "loss": 1.6958,
      "step": 151
    },
    {
      "epoch": 0.010000493445400266,
      "grad_norm": 0.7061877250671387,
      "learning_rate": 9.754288956503736e-05,
      "loss": 1.7587,
      "step": 152
    },
    {
      "epoch": 0.010066286165435794,
      "grad_norm": 0.680920422077179,
      "learning_rate": 9.750962703335547e-05,
      "loss": 1.81,
      "step": 153
    },
    {
      "epoch": 0.010132078885471322,
      "grad_norm": 0.6718249917030334,
      "learning_rate": 9.747614661986437e-05,
      "loss": 1.8675,
      "step": 154
    },
    {
      "epoch": 0.01019787160550685,
      "grad_norm": 0.6467337608337402,
      "learning_rate": 9.744244847810716e-05,
      "loss": 1.7655,
      "step": 155
    },
    {
      "epoch": 0.010263664325542378,
      "grad_norm": 0.6753641963005066,
      "learning_rate": 9.740853276262534e-05,
      "loss": 1.6277,
      "step": 156
    },
    {
      "epoch": 0.010329457045577906,
      "grad_norm": 0.5853750705718994,
      "learning_rate": 9.737439962895832e-05,
      "loss": 1.7043,
      "step": 157
    },
    {
      "epoch": 0.010395249765613434,
      "grad_norm": 0.6154720187187195,
      "learning_rate": 9.734004923364257e-05,
      "loss": 1.6254,
      "step": 158
    },
    {
      "epoch": 0.010461042485648962,
      "grad_norm": 0.6213800311088562,
      "learning_rate": 9.73054817342109e-05,
      "loss": 1.7266,
      "step": 159
    },
    {
      "epoch": 0.01052683520568449,
      "grad_norm": 0.6310060620307922,
      "learning_rate": 9.727069728919181e-05,
      "loss": 1.5622,
      "step": 160
    },
    {
      "epoch": 0.010592627925720018,
      "grad_norm": 0.6443656086921692,
      "learning_rate": 9.723569605810871e-05,
      "loss": 1.7595,
      "step": 161
    },
    {
      "epoch": 0.010658420645755546,
      "grad_norm": 0.6360974907875061,
      "learning_rate": 9.720047820147919e-05,
      "loss": 1.5975,
      "step": 162
    },
    {
      "epoch": 0.010724213365791076,
      "grad_norm": 0.648723840713501,
      "learning_rate": 9.716504388081436e-05,
      "loss": 1.6944,
      "step": 163
    },
    {
      "epoch": 0.010790006085826604,
      "grad_norm": 0.632750928401947,
      "learning_rate": 9.712939325861794e-05,
      "loss": 1.7593,
      "step": 164
    },
    {
      "epoch": 0.010855798805862132,
      "grad_norm": 0.6332002282142639,
      "learning_rate": 9.709352649838573e-05,
      "loss": 1.6756,
      "step": 165
    },
    {
      "epoch": 0.01092159152589766,
      "grad_norm": 0.6480251550674438,
      "learning_rate": 9.705744376460464e-05,
      "loss": 1.7099,
      "step": 166
    },
    {
      "epoch": 0.010987384245933188,
      "grad_norm": 0.6388484239578247,
      "learning_rate": 9.702114522275216e-05,
      "loss": 1.7719,
      "step": 167
    },
    {
      "epoch": 0.011053176965968716,
      "grad_norm": 0.5917187333106995,
      "learning_rate": 9.698463103929542e-05,
      "loss": 1.5279,
      "step": 168
    },
    {
      "epoch": 0.011118969686004244,
      "grad_norm": 0.6601297855377197,
      "learning_rate": 9.694790138169051e-05,
      "loss": 1.86,
      "step": 169
    },
    {
      "epoch": 0.011184762406039772,
      "grad_norm": 0.6304008364677429,
      "learning_rate": 9.691095641838169e-05,
      "loss": 1.6616,
      "step": 170
    },
    {
      "epoch": 0.0112505551260753,
      "grad_norm": 0.7166048288345337,
      "learning_rate": 9.687379631880062e-05,
      "loss": 1.6271,
      "step": 171
    },
    {
      "epoch": 0.011316347846110828,
      "grad_norm": 0.6489748954772949,
      "learning_rate": 9.683642125336562e-05,
      "loss": 1.6032,
      "step": 172
    },
    {
      "epoch": 0.011382140566146356,
      "grad_norm": 0.5979589223861694,
      "learning_rate": 9.679883139348082e-05,
      "loss": 1.6637,
      "step": 173
    },
    {
      "epoch": 0.011447933286181884,
      "grad_norm": 0.6470814347267151,
      "learning_rate": 9.676102691153542e-05,
      "loss": 1.5793,
      "step": 174
    },
    {
      "epoch": 0.011513726006217412,
      "grad_norm": 0.6679474711418152,
      "learning_rate": 9.67230079809029e-05,
      "loss": 1.5953,
      "step": 175
    },
    {
      "epoch": 0.01157951872625294,
      "grad_norm": 0.6367717385292053,
      "learning_rate": 9.66847747759402e-05,
      "loss": 1.5554,
      "step": 176
    },
    {
      "epoch": 0.011645311446288468,
      "grad_norm": 0.6821426749229431,
      "learning_rate": 9.664632747198693e-05,
      "loss": 1.6466,
      "step": 177
    },
    {
      "epoch": 0.011711104166323996,
      "grad_norm": 0.6893940567970276,
      "learning_rate": 9.660766624536459e-05,
      "loss": 1.6491,
      "step": 178
    },
    {
      "epoch": 0.011776896886359524,
      "grad_norm": 0.6703565716743469,
      "learning_rate": 9.656879127337571e-05,
      "loss": 1.664,
      "step": 179
    },
    {
      "epoch": 0.011842689606395052,
      "grad_norm": 0.739335298538208,
      "learning_rate": 9.652970273430312e-05,
      "loss": 1.6553,
      "step": 180
    },
    {
      "epoch": 0.01190848232643058,
      "grad_norm": 0.7087481617927551,
      "learning_rate": 9.649040080740902e-05,
      "loss": 1.6274,
      "step": 181
    },
    {
      "epoch": 0.011974275046466108,
      "grad_norm": 0.7415068745613098,
      "learning_rate": 9.645088567293426e-05,
      "loss": 1.6109,
      "step": 182
    },
    {
      "epoch": 0.012040067766501636,
      "grad_norm": 0.7282552123069763,
      "learning_rate": 9.641115751209746e-05,
      "loss": 1.5478,
      "step": 183
    },
    {
      "epoch": 0.012105860486537164,
      "grad_norm": 0.7318709492683411,
      "learning_rate": 9.637121650709417e-05,
      "loss": 1.7496,
      "step": 184
    },
    {
      "epoch": 0.012171653206572692,
      "grad_norm": 0.7077505588531494,
      "learning_rate": 9.63310628410961e-05,
      "loss": 1.697,
      "step": 185
    },
    {
      "epoch": 0.01223744592660822,
      "grad_norm": 0.7727397680282593,
      "learning_rate": 9.629069669825021e-05,
      "loss": 1.6696,
      "step": 186
    },
    {
      "epoch": 0.012303238646643748,
      "grad_norm": 0.7243698239326477,
      "learning_rate": 9.625011826367787e-05,
      "loss": 1.7203,
      "step": 187
    },
    {
      "epoch": 0.012369031366679276,
      "grad_norm": 0.754132091999054,
      "learning_rate": 9.620932772347408e-05,
      "loss": 1.6186,
      "step": 188
    },
    {
      "epoch": 0.012434824086714804,
      "grad_norm": 0.744714617729187,
      "learning_rate": 9.61683252647065e-05,
      "loss": 1.6537,
      "step": 189
    },
    {
      "epoch": 0.012500616806750332,
      "grad_norm": 0.8108708262443542,
      "learning_rate": 9.612711107541474e-05,
      "loss": 1.6096,
      "step": 190
    },
    {
      "epoch": 0.01256640952678586,
      "grad_norm": 0.7732747793197632,
      "learning_rate": 9.608568534460936e-05,
      "loss": 1.5707,
      "step": 191
    },
    {
      "epoch": 0.01263220224682139,
      "grad_norm": 0.8782806396484375,
      "learning_rate": 9.60440482622711e-05,
      "loss": 1.6991,
      "step": 192
    },
    {
      "epoch": 0.012697994966856918,
      "grad_norm": 0.8542294502258301,
      "learning_rate": 9.600220001934993e-05,
      "loss": 1.4367,
      "step": 193
    },
    {
      "epoch": 0.012763787686892446,
      "grad_norm": 0.877135694026947,
      "learning_rate": 9.596014080776423e-05,
      "loss": 1.6514,
      "step": 194
    },
    {
      "epoch": 0.012829580406927974,
      "grad_norm": 0.9459388256072998,
      "learning_rate": 9.591787082039992e-05,
      "loss": 1.6164,
      "step": 195
    },
    {
      "epoch": 0.012895373126963502,
      "grad_norm": 0.8651464581489563,
      "learning_rate": 9.587539025110952e-05,
      "loss": 1.6191,
      "step": 196
    },
    {
      "epoch": 0.01296116584699903,
      "grad_norm": 0.8998655080795288,
      "learning_rate": 9.583269929471128e-05,
      "loss": 1.4867,
      "step": 197
    },
    {
      "epoch": 0.013026958567034558,
      "grad_norm": 0.9345824122428894,
      "learning_rate": 9.578979814698835e-05,
      "loss": 1.6862,
      "step": 198
    },
    {
      "epoch": 0.013092751287070086,
      "grad_norm": 1.0395649671554565,
      "learning_rate": 9.574668700468777e-05,
      "loss": 1.4863,
      "step": 199
    },
    {
      "epoch": 0.013158544007105614,
      "grad_norm": 1.145094871520996,
      "learning_rate": 9.570336606551967e-05,
      "loss": 1.534,
      "step": 200
    },
    {
      "epoch": 0.013224336727141142,
      "grad_norm": 0.6736162304878235,
      "learning_rate": 9.565983552815628e-05,
      "loss": 1.7361,
      "step": 201
    },
    {
      "epoch": 0.01329012944717667,
      "grad_norm": 0.7287967205047607,
      "learning_rate": 9.56160955922311e-05,
      "loss": 1.9231,
      "step": 202
    },
    {
      "epoch": 0.013355922167212198,
      "grad_norm": 0.6452571153640747,
      "learning_rate": 9.557214645833792e-05,
      "loss": 1.7251,
      "step": 203
    },
    {
      "epoch": 0.013421714887247726,
      "grad_norm": 0.6660079956054688,
      "learning_rate": 9.552798832802993e-05,
      "loss": 1.7163,
      "step": 204
    },
    {
      "epoch": 0.013487507607283254,
      "grad_norm": 0.6526784896850586,
      "learning_rate": 9.548362140381876e-05,
      "loss": 1.6819,
      "step": 205
    },
    {
      "epoch": 0.013553300327318782,
      "grad_norm": 0.6540437340736389,
      "learning_rate": 9.543904588917367e-05,
      "loss": 1.7811,
      "step": 206
    },
    {
      "epoch": 0.01361909304735431,
      "grad_norm": 0.5529007911682129,
      "learning_rate": 9.53942619885204e-05,
      "loss": 1.6977,
      "step": 207
    },
    {
      "epoch": 0.013684885767389838,
      "grad_norm": 0.608326256275177,
      "learning_rate": 9.534926990724046e-05,
      "loss": 1.592,
      "step": 208
    },
    {
      "epoch": 0.013750678487425366,
      "grad_norm": 0.6321333050727844,
      "learning_rate": 9.530406985167004e-05,
      "loss": 1.7665,
      "step": 209
    },
    {
      "epoch": 0.013816471207460894,
      "grad_norm": 0.592158317565918,
      "learning_rate": 9.525866202909915e-05,
      "loss": 1.722,
      "step": 210
    },
    {
      "epoch": 0.013882263927496422,
      "grad_norm": 0.5666524767875671,
      "learning_rate": 9.521304664777058e-05,
      "loss": 1.652,
      "step": 211
    },
    {
      "epoch": 0.01394805664753195,
      "grad_norm": 0.6615280508995056,
      "learning_rate": 9.516722391687902e-05,
      "loss": 1.8336,
      "step": 212
    },
    {
      "epoch": 0.014013849367567478,
      "grad_norm": 0.6410874128341675,
      "learning_rate": 9.51211940465701e-05,
      "loss": 1.8056,
      "step": 213
    },
    {
      "epoch": 0.014079642087603006,
      "grad_norm": 0.6361430287361145,
      "learning_rate": 9.507495724793937e-05,
      "loss": 1.8389,
      "step": 214
    },
    {
      "epoch": 0.014145434807638534,
      "grad_norm": 0.6516650915145874,
      "learning_rate": 9.502851373303136e-05,
      "loss": 1.7175,
      "step": 215
    },
    {
      "epoch": 0.014211227527674062,
      "grad_norm": 2.113678455352783,
      "learning_rate": 9.498186371483868e-05,
      "loss": 1.6789,
      "step": 216
    },
    {
      "epoch": 0.01427702024770959,
      "grad_norm": 0.6748088598251343,
      "learning_rate": 9.493500740730089e-05,
      "loss": 1.6288,
      "step": 217
    },
    {
      "epoch": 0.014342812967745118,
      "grad_norm": 0.6387640833854675,
      "learning_rate": 9.488794502530362e-05,
      "loss": 1.6993,
      "step": 218
    },
    {
      "epoch": 0.014408605687780646,
      "grad_norm": 0.7229827046394348,
      "learning_rate": 9.484067678467761e-05,
      "loss": 1.5233,
      "step": 219
    },
    {
      "epoch": 0.014474398407816174,
      "grad_norm": 0.6891183257102966,
      "learning_rate": 9.479320290219766e-05,
      "loss": 1.7135,
      "step": 220
    },
    {
      "epoch": 0.014540191127851702,
      "grad_norm": 0.6379052400588989,
      "learning_rate": 9.474552359558166e-05,
      "loss": 1.5541,
      "step": 221
    },
    {
      "epoch": 0.014605983847887232,
      "grad_norm": 0.7389596104621887,
      "learning_rate": 9.469763908348957e-05,
      "loss": 1.7669,
      "step": 222
    },
    {
      "epoch": 0.01467177656792276,
      "grad_norm": 0.6853717565536499,
      "learning_rate": 9.464954958552244e-05,
      "loss": 1.6764,
      "step": 223
    },
    {
      "epoch": 0.014737569287958288,
      "grad_norm": 0.6811705231666565,
      "learning_rate": 9.460125532222141e-05,
      "loss": 1.6761,
      "step": 224
    },
    {
      "epoch": 0.014803362007993816,
      "grad_norm": 0.7454027533531189,
      "learning_rate": 9.455275651506665e-05,
      "loss": 1.657,
      "step": 225
    },
    {
      "epoch": 0.014869154728029344,
      "grad_norm": 0.6707807779312134,
      "learning_rate": 9.450405338647645e-05,
      "loss": 1.6695,
      "step": 226
    },
    {
      "epoch": 0.014934947448064872,
      "grad_norm": 0.698140025138855,
      "learning_rate": 9.445514615980604e-05,
      "loss": 1.658,
      "step": 227
    },
    {
      "epoch": 0.0150007401681004,
      "grad_norm": 0.6676850318908691,
      "learning_rate": 9.44060350593467e-05,
      "loss": 1.5562,
      "step": 228
    },
    {
      "epoch": 0.015066532888135928,
      "grad_norm": 0.7539533972740173,
      "learning_rate": 9.435672031032474e-05,
      "loss": 1.604,
      "step": 229
    },
    {
      "epoch": 0.015132325608171456,
      "grad_norm": 0.7509004473686218,
      "learning_rate": 9.43072021389003e-05,
      "loss": 1.5699,
      "step": 230
    },
    {
      "epoch": 0.015198118328206984,
      "grad_norm": 0.7379850149154663,
      "learning_rate": 9.425748077216649e-05,
      "loss": 1.6537,
      "step": 231
    },
    {
      "epoch": 0.015263911048242512,
      "grad_norm": 0.7617546319961548,
      "learning_rate": 9.420755643814833e-05,
      "loss": 1.7029,
      "step": 232
    },
    {
      "epoch": 0.01532970376827804,
      "grad_norm": 0.7647762894630432,
      "learning_rate": 9.415742936580157e-05,
      "loss": 1.6908,
      "step": 233
    },
    {
      "epoch": 0.015395496488313568,
      "grad_norm": 0.7109963297843933,
      "learning_rate": 9.410709978501177e-05,
      "loss": 1.6588,
      "step": 234
    },
    {
      "epoch": 0.015461289208349096,
      "grad_norm": 0.793974757194519,
      "learning_rate": 9.405656792659321e-05,
      "loss": 1.7287,
      "step": 235
    },
    {
      "epoch": 0.015527081928384624,
      "grad_norm": 0.7597332000732422,
      "learning_rate": 9.400583402228784e-05,
      "loss": 1.721,
      "step": 236
    },
    {
      "epoch": 0.015592874648420152,
      "grad_norm": 0.8103028535842896,
      "learning_rate": 9.395489830476417e-05,
      "loss": 1.5775,
      "step": 237
    },
    {
      "epoch": 0.01565866736845568,
      "grad_norm": 0.738635241985321,
      "learning_rate": 9.390376100761624e-05,
      "loss": 1.4594,
      "step": 238
    },
    {
      "epoch": 0.01572446008849121,
      "grad_norm": 0.7879254817962646,
      "learning_rate": 9.38524223653626e-05,
      "loss": 1.7778,
      "step": 239
    },
    {
      "epoch": 0.015790252808526736,
      "grad_norm": 0.7641370892524719,
      "learning_rate": 9.380088261344509e-05,
      "loss": 1.6935,
      "step": 240
    },
    {
      "epoch": 0.015856045528562266,
      "grad_norm": 0.8217648863792419,
      "learning_rate": 9.374914198822791e-05,
      "loss": 1.6218,
      "step": 241
    },
    {
      "epoch": 0.015921838248597792,
      "grad_norm": 0.799241840839386,
      "learning_rate": 9.369720072699647e-05,
      "loss": 1.6173,
      "step": 242
    },
    {
      "epoch": 0.015987630968633322,
      "grad_norm": 0.7930746674537659,
      "learning_rate": 9.364505906795631e-05,
      "loss": 1.6816,
      "step": 243
    },
    {
      "epoch": 0.01605342368866885,
      "grad_norm": 0.9224853515625,
      "learning_rate": 9.359271725023198e-05,
      "loss": 1.8139,
      "step": 244
    },
    {
      "epoch": 0.016119216408704378,
      "grad_norm": 0.8082649111747742,
      "learning_rate": 9.354017551386599e-05,
      "loss": 1.5355,
      "step": 245
    },
    {
      "epoch": 0.016185009128739904,
      "grad_norm": 0.9044872522354126,
      "learning_rate": 9.34874340998177e-05,
      "loss": 1.5935,
      "step": 246
    },
    {
      "epoch": 0.016250801848775434,
      "grad_norm": 0.8898908495903015,
      "learning_rate": 9.343449324996217e-05,
      "loss": 1.6187,
      "step": 247
    },
    {
      "epoch": 0.01631659456881096,
      "grad_norm": 0.8817726969718933,
      "learning_rate": 9.338135320708911e-05,
      "loss": 1.6481,
      "step": 248
    },
    {
      "epoch": 0.01638238728884649,
      "grad_norm": 1.0543079376220703,
      "learning_rate": 9.332801421490174e-05,
      "loss": 1.6283,
      "step": 249
    },
    {
      "epoch": 0.016448180008882016,
      "grad_norm": 1.1242119073867798,
      "learning_rate": 9.327447651801564e-05,
      "loss": 1.4045,
      "step": 250
    },
    {
      "epoch": 0.016513972728917546,
      "grad_norm": 0.6645945310592651,
      "learning_rate": 9.322074036195769e-05,
      "loss": 1.7638,
      "step": 251
    },
    {
      "epoch": 0.016579765448953072,
      "grad_norm": 0.636358380317688,
      "learning_rate": 9.316680599316493e-05,
      "loss": 1.8369,
      "step": 252
    },
    {
      "epoch": 0.016645558168988602,
      "grad_norm": 0.6823999881744385,
      "learning_rate": 9.311267365898337e-05,
      "loss": 1.7382,
      "step": 253
    },
    {
      "epoch": 0.01671135088902413,
      "grad_norm": 0.6345416903495789,
      "learning_rate": 9.305834360766695e-05,
      "loss": 1.6073,
      "step": 254
    },
    {
      "epoch": 0.016777143609059658,
      "grad_norm": 0.6101424694061279,
      "learning_rate": 9.300381608837631e-05,
      "loss": 1.705,
      "step": 255
    },
    {
      "epoch": 0.016842936329095184,
      "grad_norm": 0.6029302477836609,
      "learning_rate": 9.294909135117771e-05,
      "loss": 1.8328,
      "step": 256
    },
    {
      "epoch": 0.016908729049130714,
      "grad_norm": 0.5847995281219482,
      "learning_rate": 9.289416964704185e-05,
      "loss": 1.6327,
      "step": 257
    },
    {
      "epoch": 0.01697452176916624,
      "grad_norm": 0.6195299625396729,
      "learning_rate": 9.283905122784277e-05,
      "loss": 1.7127,
      "step": 258
    },
    {
      "epoch": 0.01704031448920177,
      "grad_norm": 0.5608875751495361,
      "learning_rate": 9.278373634635659e-05,
      "loss": 1.5031,
      "step": 259
    },
    {
      "epoch": 0.017106107209237296,
      "grad_norm": 0.5664215683937073,
      "learning_rate": 9.272822525626046e-05,
      "loss": 1.6431,
      "step": 260
    },
    {
      "epoch": 0.017171899929272826,
      "grad_norm": 0.5837283134460449,
      "learning_rate": 9.267251821213137e-05,
      "loss": 1.6068,
      "step": 261
    },
    {
      "epoch": 0.017237692649308353,
      "grad_norm": 0.6226207613945007,
      "learning_rate": 9.261661546944491e-05,
      "loss": 1.6153,
      "step": 262
    },
    {
      "epoch": 0.017303485369343882,
      "grad_norm": 0.6285345554351807,
      "learning_rate": 9.25605172845742e-05,
      "loss": 1.6145,
      "step": 263
    },
    {
      "epoch": 0.01736927808937941,
      "grad_norm": 0.6322391629219055,
      "learning_rate": 9.250422391478868e-05,
      "loss": 1.7426,
      "step": 264
    },
    {
      "epoch": 0.01743507080941494,
      "grad_norm": 0.599488377571106,
      "learning_rate": 9.244773561825287e-05,
      "loss": 1.5602,
      "step": 265
    },
    {
      "epoch": 0.017500863529450468,
      "grad_norm": 0.6227719187736511,
      "learning_rate": 9.239105265402525e-05,
      "loss": 1.5007,
      "step": 266
    },
    {
      "epoch": 0.017566656249485994,
      "grad_norm": 0.6282662153244019,
      "learning_rate": 9.23341752820571e-05,
      "loss": 1.6923,
      "step": 267
    },
    {
      "epoch": 0.017632448969521524,
      "grad_norm": 0.6106943488121033,
      "learning_rate": 9.22771037631912e-05,
      "loss": 1.5653,
      "step": 268
    },
    {
      "epoch": 0.01769824168955705,
      "grad_norm": 0.6486127972602844,
      "learning_rate": 9.221983835916074e-05,
      "loss": 1.568,
      "step": 269
    },
    {
      "epoch": 0.01776403440959258,
      "grad_norm": 0.6146793365478516,
      "learning_rate": 9.21623793325881e-05,
      "loss": 1.6671,
      "step": 270
    },
    {
      "epoch": 0.017829827129628106,
      "grad_norm": 0.6328508257865906,
      "learning_rate": 9.210472694698355e-05,
      "loss": 1.6419,
      "step": 271
    },
    {
      "epoch": 0.017895619849663636,
      "grad_norm": 0.6605221033096313,
      "learning_rate": 9.204688146674418e-05,
      "loss": 1.7364,
      "step": 272
    },
    {
      "epoch": 0.017961412569699162,
      "grad_norm": 0.6620165705680847,
      "learning_rate": 9.198884315715259e-05,
      "loss": 1.6665,
      "step": 273
    },
    {
      "epoch": 0.018027205289734692,
      "grad_norm": 0.6743021011352539,
      "learning_rate": 9.193061228437572e-05,
      "loss": 1.5406,
      "step": 274
    },
    {
      "epoch": 0.01809299800977022,
      "grad_norm": 0.6547907590866089,
      "learning_rate": 9.187218911546362e-05,
      "loss": 1.5747,
      "step": 275
    },
    {
      "epoch": 0.018158790729805748,
      "grad_norm": 0.7358336448669434,
      "learning_rate": 9.18135739183482e-05,
      "loss": 1.6917,
      "step": 276
    },
    {
      "epoch": 0.018224583449841274,
      "grad_norm": 0.6380569338798523,
      "learning_rate": 9.175476696184205e-05,
      "loss": 1.5399,
      "step": 277
    },
    {
      "epoch": 0.018290376169876804,
      "grad_norm": 0.7086791396141052,
      "learning_rate": 9.169576851563715e-05,
      "loss": 1.6781,
      "step": 278
    },
    {
      "epoch": 0.01835616888991233,
      "grad_norm": 0.7136445641517639,
      "learning_rate": 9.163657885030368e-05,
      "loss": 1.611,
      "step": 279
    },
    {
      "epoch": 0.01842196160994786,
      "grad_norm": 0.6863766312599182,
      "learning_rate": 9.157719823728876e-05,
      "loss": 1.6871,
      "step": 280
    },
    {
      "epoch": 0.018487754329983386,
      "grad_norm": 0.7350311875343323,
      "learning_rate": 9.151762694891521e-05,
      "loss": 1.508,
      "step": 281
    },
    {
      "epoch": 0.018553547050018916,
      "grad_norm": 0.6730437874794006,
      "learning_rate": 9.14578652583803e-05,
      "loss": 1.5196,
      "step": 282
    },
    {
      "epoch": 0.018619339770054442,
      "grad_norm": 0.718139111995697,
      "learning_rate": 9.139791343975448e-05,
      "loss": 1.5038,
      "step": 283
    },
    {
      "epoch": 0.018685132490089972,
      "grad_norm": 0.7707210183143616,
      "learning_rate": 9.133777176798013e-05,
      "loss": 1.729,
      "step": 284
    },
    {
      "epoch": 0.0187509252101255,
      "grad_norm": 0.7235831618309021,
      "learning_rate": 9.127744051887035e-05,
      "loss": 1.565,
      "step": 285
    },
    {
      "epoch": 0.018816717930161028,
      "grad_norm": 0.7182489633560181,
      "learning_rate": 9.121691996910762e-05,
      "loss": 1.6169,
      "step": 286
    },
    {
      "epoch": 0.018882510650196555,
      "grad_norm": 0.7400686144828796,
      "learning_rate": 9.115621039624256e-05,
      "loss": 1.653,
      "step": 287
    },
    {
      "epoch": 0.018948303370232084,
      "grad_norm": 0.7060942649841309,
      "learning_rate": 9.109531207869266e-05,
      "loss": 1.7122,
      "step": 288
    },
    {
      "epoch": 0.01901409609026761,
      "grad_norm": 0.7848304510116577,
      "learning_rate": 9.103422529574104e-05,
      "loss": 1.7381,
      "step": 289
    },
    {
      "epoch": 0.01907988881030314,
      "grad_norm": 0.7508149743080139,
      "learning_rate": 9.09729503275351e-05,
      "loss": 1.6243,
      "step": 290
    },
    {
      "epoch": 0.019145681530338667,
      "grad_norm": 0.765107274055481,
      "learning_rate": 9.091148745508526e-05,
      "loss": 1.6244,
      "step": 291
    },
    {
      "epoch": 0.019211474250374196,
      "grad_norm": 0.8006464242935181,
      "learning_rate": 9.08498369602637e-05,
      "loss": 1.5571,
      "step": 292
    },
    {
      "epoch": 0.019277266970409723,
      "grad_norm": 0.7692310214042664,
      "learning_rate": 9.078799912580304e-05,
      "loss": 1.6686,
      "step": 293
    },
    {
      "epoch": 0.019343059690445252,
      "grad_norm": 0.8822973966598511,
      "learning_rate": 9.072597423529508e-05,
      "loss": 1.6505,
      "step": 294
    },
    {
      "epoch": 0.01940885241048078,
      "grad_norm": 0.8218017220497131,
      "learning_rate": 9.066376257318938e-05,
      "loss": 1.5566,
      "step": 295
    },
    {
      "epoch": 0.01947464513051631,
      "grad_norm": 0.8378568887710571,
      "learning_rate": 9.060136442479215e-05,
      "loss": 1.5851,
      "step": 296
    },
    {
      "epoch": 0.019540437850551838,
      "grad_norm": 0.9615349173545837,
      "learning_rate": 9.053878007626478e-05,
      "loss": 1.6833,
      "step": 297
    },
    {
      "epoch": 0.019606230570587364,
      "grad_norm": 0.8999234437942505,
      "learning_rate": 9.04760098146226e-05,
      "loss": 1.5815,
      "step": 298
    },
    {
      "epoch": 0.019672023290622894,
      "grad_norm": 1.0168418884277344,
      "learning_rate": 9.041305392773354e-05,
      "loss": 1.7269,
      "step": 299
    },
    {
      "epoch": 0.01973781601065842,
      "grad_norm": 1.2946468591690063,
      "learning_rate": 9.034991270431681e-05,
      "loss": 1.6278,
      "step": 300
    },
    {
      "epoch": 0.01980360873069395,
      "grad_norm": 0.6013835668563843,
      "learning_rate": 9.02865864339416e-05,
      "loss": 1.7835,
      "step": 301
    },
    {
      "epoch": 0.019869401450729476,
      "grad_norm": 0.6507825255393982,
      "learning_rate": 9.022307540702576e-05,
      "loss": 1.8789,
      "step": 302
    },
    {
      "epoch": 0.019935194170765006,
      "grad_norm": 0.6571053862571716,
      "learning_rate": 9.015937991483439e-05,
      "loss": 1.8656,
      "step": 303
    },
    {
      "epoch": 0.020000986890800532,
      "grad_norm": 0.669532060623169,
      "learning_rate": 9.009550024947856e-05,
      "loss": 1.7856,
      "step": 304
    },
    {
      "epoch": 0.020066779610836062,
      "grad_norm": 0.6802514791488647,
      "learning_rate": 9.003143670391403e-05,
      "loss": 1.8545,
      "step": 305
    },
    {
      "epoch": 0.02013257233087159,
      "grad_norm": 0.6057747602462769,
      "learning_rate": 8.996718957193978e-05,
      "loss": 1.6657,
      "step": 306
    },
    {
      "epoch": 0.020198365050907118,
      "grad_norm": 0.5737309455871582,
      "learning_rate": 8.990275914819679e-05,
      "loss": 1.7686,
      "step": 307
    },
    {
      "epoch": 0.020264157770942644,
      "grad_norm": 0.5848949551582336,
      "learning_rate": 8.983814572816656e-05,
      "loss": 1.7321,
      "step": 308
    },
    {
      "epoch": 0.020329950490978174,
      "grad_norm": 0.5747925043106079,
      "learning_rate": 8.977334960816986e-05,
      "loss": 1.6342,
      "step": 309
    },
    {
      "epoch": 0.0203957432110137,
      "grad_norm": 0.6126972436904907,
      "learning_rate": 8.970837108536532e-05,
      "loss": 1.5879,
      "step": 310
    },
    {
      "epoch": 0.02046153593104923,
      "grad_norm": 0.6280604600906372,
      "learning_rate": 8.964321045774807e-05,
      "loss": 1.8228,
      "step": 311
    },
    {
      "epoch": 0.020527328651084756,
      "grad_norm": 0.6287879347801208,
      "learning_rate": 8.957786802414842e-05,
      "loss": 1.6301,
      "step": 312
    },
    {
      "epoch": 0.020593121371120286,
      "grad_norm": 0.6327916383743286,
      "learning_rate": 8.951234408423042e-05,
      "loss": 1.8064,
      "step": 313
    },
    {
      "epoch": 0.020658914091155813,
      "grad_norm": 0.6169136166572571,
      "learning_rate": 8.944663893849052e-05,
      "loss": 1.627,
      "step": 314
    },
    {
      "epoch": 0.020724706811191342,
      "grad_norm": 0.7315105199813843,
      "learning_rate": 8.938075288825622e-05,
      "loss": 1.5807,
      "step": 315
    },
    {
      "epoch": 0.02079049953122687,
      "grad_norm": 0.6365809440612793,
      "learning_rate": 8.93146862356846e-05,
      "loss": 1.775,
      "step": 316
    },
    {
      "epoch": 0.0208562922512624,
      "grad_norm": 0.6544710993766785,
      "learning_rate": 8.924843928376104e-05,
      "loss": 1.7399,
      "step": 317
    },
    {
      "epoch": 0.020922084971297925,
      "grad_norm": 0.5984869599342346,
      "learning_rate": 8.91820123362978e-05,
      "loss": 1.5659,
      "step": 318
    },
    {
      "epoch": 0.020987877691333454,
      "grad_norm": 0.6417421698570251,
      "learning_rate": 8.911540569793253e-05,
      "loss": 1.8156,
      "step": 319
    },
    {
      "epoch": 0.02105367041136898,
      "grad_norm": 0.6159281134605408,
      "learning_rate": 8.904861967412703e-05,
      "loss": 1.6519,
      "step": 320
    },
    {
      "epoch": 0.02111946313140451,
      "grad_norm": 0.6258522868156433,
      "learning_rate": 8.898165457116574e-05,
      "loss": 1.6219,
      "step": 321
    },
    {
      "epoch": 0.021185255851440037,
      "grad_norm": 0.6070511937141418,
      "learning_rate": 8.891451069615437e-05,
      "loss": 1.5032,
      "step": 322
    },
    {
      "epoch": 0.021251048571475566,
      "grad_norm": 0.6364402770996094,
      "learning_rate": 8.884718835701848e-05,
      "loss": 1.5849,
      "step": 323
    },
    {
      "epoch": 0.021316841291511093,
      "grad_norm": 0.6753417253494263,
      "learning_rate": 8.877968786250212e-05,
      "loss": 1.6483,
      "step": 324
    },
    {
      "epoch": 0.021382634011546622,
      "grad_norm": 0.6451675295829773,
      "learning_rate": 8.87120095221663e-05,
      "loss": 1.6688,
      "step": 325
    },
    {
      "epoch": 0.021448426731582152,
      "grad_norm": 0.6772330403327942,
      "learning_rate": 8.86441536463877e-05,
      "loss": 1.6518,
      "step": 326
    },
    {
      "epoch": 0.02151421945161768,
      "grad_norm": 0.6775709986686707,
      "learning_rate": 8.857612054635713e-05,
      "loss": 1.6329,
      "step": 327
    },
    {
      "epoch": 0.021580012171653208,
      "grad_norm": 0.7176543474197388,
      "learning_rate": 8.850791053407824e-05,
      "loss": 1.7176,
      "step": 328
    },
    {
      "epoch": 0.021645804891688734,
      "grad_norm": 0.8762123584747314,
      "learning_rate": 8.843952392236594e-05,
      "loss": 1.7178,
      "step": 329
    },
    {
      "epoch": 0.021711597611724264,
      "grad_norm": 0.680343508720398,
      "learning_rate": 8.837096102484508e-05,
      "loss": 1.7253,
      "step": 330
    },
    {
      "epoch": 0.02177739033175979,
      "grad_norm": 0.6655333042144775,
      "learning_rate": 8.83022221559489e-05,
      "loss": 1.5397,
      "step": 331
    },
    {
      "epoch": 0.02184318305179532,
      "grad_norm": 0.7247670292854309,
      "learning_rate": 8.823330763091775e-05,
      "loss": 1.5323,
      "step": 332
    },
    {
      "epoch": 0.021908975771830846,
      "grad_norm": 0.6783066987991333,
      "learning_rate": 8.816421776579749e-05,
      "loss": 1.578,
      "step": 333
    },
    {
      "epoch": 0.021974768491866376,
      "grad_norm": 0.7266553640365601,
      "learning_rate": 8.80949528774381e-05,
      "loss": 1.449,
      "step": 334
    },
    {
      "epoch": 0.022040561211901902,
      "grad_norm": 0.7339120507240295,
      "learning_rate": 8.802551328349222e-05,
      "loss": 1.5265,
      "step": 335
    },
    {
      "epoch": 0.022106353931937432,
      "grad_norm": 0.703031063079834,
      "learning_rate": 8.795589930241374e-05,
      "loss": 1.4054,
      "step": 336
    },
    {
      "epoch": 0.02217214665197296,
      "grad_norm": 0.8293249011039734,
      "learning_rate": 8.788611125345625e-05,
      "loss": 1.613,
      "step": 337
    },
    {
      "epoch": 0.022237939372008488,
      "grad_norm": 0.8584293127059937,
      "learning_rate": 8.781614945667169e-05,
      "loss": 1.693,
      "step": 338
    },
    {
      "epoch": 0.022303732092044014,
      "grad_norm": 0.7650991082191467,
      "learning_rate": 8.774601423290875e-05,
      "loss": 1.6419,
      "step": 339
    },
    {
      "epoch": 0.022369524812079544,
      "grad_norm": 0.8328055143356323,
      "learning_rate": 8.767570590381148e-05,
      "loss": 1.6992,
      "step": 340
    },
    {
      "epoch": 0.02243531753211507,
      "grad_norm": 0.8707652688026428,
      "learning_rate": 8.760522479181784e-05,
      "loss": 1.8145,
      "step": 341
    },
    {
      "epoch": 0.0225011102521506,
      "grad_norm": 0.8000150918960571,
      "learning_rate": 8.753457122015812e-05,
      "loss": 1.6967,
      "step": 342
    },
    {
      "epoch": 0.022566902972186127,
      "grad_norm": 0.7786499261856079,
      "learning_rate": 8.746374551285358e-05,
      "loss": 1.5119,
      "step": 343
    },
    {
      "epoch": 0.022632695692221656,
      "grad_norm": 0.8160056471824646,
      "learning_rate": 8.73927479947149e-05,
      "loss": 1.451,
      "step": 344
    },
    {
      "epoch": 0.022698488412257183,
      "grad_norm": 0.8968502283096313,
      "learning_rate": 8.732157899134063e-05,
      "loss": 1.5002,
      "step": 345
    },
    {
      "epoch": 0.022764281132292712,
      "grad_norm": 0.774455726146698,
      "learning_rate": 8.725023882911583e-05,
      "loss": 1.6271,
      "step": 346
    },
    {
      "epoch": 0.02283007385232824,
      "grad_norm": 0.8428329229354858,
      "learning_rate": 8.717872783521047e-05,
      "loss": 1.5726,
      "step": 347
    },
    {
      "epoch": 0.02289586657236377,
      "grad_norm": 0.972919225692749,
      "learning_rate": 8.710704633757796e-05,
      "loss": 1.4679,
      "step": 348
    },
    {
      "epoch": 0.022961659292399295,
      "grad_norm": 1.157837152481079,
      "learning_rate": 8.70351946649537e-05,
      "loss": 1.4656,
      "step": 349
    },
    {
      "epoch": 0.023027452012434824,
      "grad_norm": 1.2611714601516724,
      "learning_rate": 8.696317314685341e-05,
      "loss": 1.428,
      "step": 350
    },
    {
      "epoch": 0.02309324473247035,
      "grad_norm": 0.6036838889122009,
      "learning_rate": 8.689098211357187e-05,
      "loss": 1.7917,
      "step": 351
    },
    {
      "epoch": 0.02315903745250588,
      "grad_norm": 0.6515650153160095,
      "learning_rate": 8.681862189618118e-05,
      "loss": 1.7356,
      "step": 352
    },
    {
      "epoch": 0.023224830172541407,
      "grad_norm": 0.643792450428009,
      "learning_rate": 8.674609282652934e-05,
      "loss": 1.8185,
      "step": 353
    },
    {
      "epoch": 0.023290622892576936,
      "grad_norm": 0.6375747323036194,
      "learning_rate": 8.667339523723875e-05,
      "loss": 1.7393,
      "step": 354
    },
    {
      "epoch": 0.023356415612612466,
      "grad_norm": 0.601500391960144,
      "learning_rate": 8.660052946170459e-05,
      "loss": 1.6424,
      "step": 355
    },
    {
      "epoch": 0.023422208332647992,
      "grad_norm": 0.584318220615387,
      "learning_rate": 8.65274958340934e-05,
      "loss": 1.6143,
      "step": 356
    },
    {
      "epoch": 0.023488001052683522,
      "grad_norm": 0.5999240875244141,
      "learning_rate": 8.645429468934147e-05,
      "loss": 1.7236,
      "step": 357
    },
    {
      "epoch": 0.02355379377271905,
      "grad_norm": 0.6196637153625488,
      "learning_rate": 8.638092636315338e-05,
      "loss": 1.7524,
      "step": 358
    },
    {
      "epoch": 0.023619586492754578,
      "grad_norm": 0.6153246164321899,
      "learning_rate": 8.630739119200035e-05,
      "loss": 1.5623,
      "step": 359
    },
    {
      "epoch": 0.023685379212790104,
      "grad_norm": 0.5839650630950928,
      "learning_rate": 8.623368951311881e-05,
      "loss": 1.6695,
      "step": 360
    },
    {
      "epoch": 0.023751171932825634,
      "grad_norm": 0.5994311571121216,
      "learning_rate": 8.615982166450878e-05,
      "loss": 1.6348,
      "step": 361
    },
    {
      "epoch": 0.02381696465286116,
      "grad_norm": 0.6017064452171326,
      "learning_rate": 8.608578798493236e-05,
      "loss": 1.6742,
      "step": 362
    },
    {
      "epoch": 0.02388275737289669,
      "grad_norm": 0.6123796105384827,
      "learning_rate": 8.601158881391212e-05,
      "loss": 1.5242,
      "step": 363
    },
    {
      "epoch": 0.023948550092932216,
      "grad_norm": 0.5843928456306458,
      "learning_rate": 8.593722449172965e-05,
      "loss": 1.5024,
      "step": 364
    },
    {
      "epoch": 0.024014342812967746,
      "grad_norm": 0.5963286757469177,
      "learning_rate": 8.586269535942385e-05,
      "loss": 1.6205,
      "step": 365
    },
    {
      "epoch": 0.024080135533003273,
      "grad_norm": 0.5980545282363892,
      "learning_rate": 8.578800175878954e-05,
      "loss": 1.4792,
      "step": 366
    },
    {
      "epoch": 0.024145928253038802,
      "grad_norm": 0.6400820016860962,
      "learning_rate": 8.571314403237572e-05,
      "loss": 1.7103,
      "step": 367
    },
    {
      "epoch": 0.02421172097307433,
      "grad_norm": 0.6095431447029114,
      "learning_rate": 8.563812252348411e-05,
      "loss": 1.4906,
      "step": 368
    },
    {
      "epoch": 0.02427751369310986,
      "grad_norm": 0.6984130144119263,
      "learning_rate": 8.556293757616757e-05,
      "loss": 1.8139,
      "step": 369
    },
    {
      "epoch": 0.024343306413145385,
      "grad_norm": 0.6309308409690857,
      "learning_rate": 8.548758953522849e-05,
      "loss": 1.6624,
      "step": 370
    },
    {
      "epoch": 0.024409099133180914,
      "grad_norm": 0.5984780788421631,
      "learning_rate": 8.541207874621718e-05,
      "loss": 1.5398,
      "step": 371
    },
    {
      "epoch": 0.02447489185321644,
      "grad_norm": 0.7127171754837036,
      "learning_rate": 8.533640555543034e-05,
      "loss": 1.8378,
      "step": 372
    },
    {
      "epoch": 0.02454068457325197,
      "grad_norm": 0.6665642857551575,
      "learning_rate": 8.526057030990947e-05,
      "loss": 1.5849,
      "step": 373
    },
    {
      "epoch": 0.024606477293287497,
      "grad_norm": 0.6169716715812683,
      "learning_rate": 8.518457335743926e-05,
      "loss": 1.6305,
      "step": 374
    },
    {
      "epoch": 0.024672270013323026,
      "grad_norm": 0.6478058099746704,
      "learning_rate": 8.510841504654596e-05,
      "loss": 1.4958,
      "step": 375
    },
    {
      "epoch": 0.024738062733358553,
      "grad_norm": 0.672910749912262,
      "learning_rate": 8.50320957264959e-05,
      "loss": 1.7338,
      "step": 376
    },
    {
      "epoch": 0.024803855453394082,
      "grad_norm": 0.7039096355438232,
      "learning_rate": 8.495561574729369e-05,
      "loss": 1.5893,
      "step": 377
    },
    {
      "epoch": 0.02486964817342961,
      "grad_norm": 0.6732673645019531,
      "learning_rate": 8.487897545968084e-05,
      "loss": 1.6202,
      "step": 378
    },
    {
      "epoch": 0.02493544089346514,
      "grad_norm": 0.6896900534629822,
      "learning_rate": 8.480217521513399e-05,
      "loss": 1.6855,
      "step": 379
    },
    {
      "epoch": 0.025001233613500665,
      "grad_norm": 0.6888880729675293,
      "learning_rate": 8.472521536586335e-05,
      "loss": 1.7113,
      "step": 380
    },
    {
      "epoch": 0.025067026333536194,
      "grad_norm": 0.7276954650878906,
      "learning_rate": 8.464809626481111e-05,
      "loss": 1.7445,
      "step": 381
    },
    {
      "epoch": 0.02513281905357172,
      "grad_norm": 0.6776707172393799,
      "learning_rate": 8.457081826564979e-05,
      "loss": 1.5922,
      "step": 382
    },
    {
      "epoch": 0.02519861177360725,
      "grad_norm": 0.6573496460914612,
      "learning_rate": 8.449338172278059e-05,
      "loss": 1.6475,
      "step": 383
    },
    {
      "epoch": 0.02526440449364278,
      "grad_norm": 0.7136342525482178,
      "learning_rate": 8.441578699133185e-05,
      "loss": 1.7994,
      "step": 384
    },
    {
      "epoch": 0.025330197213678306,
      "grad_norm": 0.702238917350769,
      "learning_rate": 8.433803442715735e-05,
      "loss": 1.5636,
      "step": 385
    },
    {
      "epoch": 0.025395989933713836,
      "grad_norm": 0.754508376121521,
      "learning_rate": 8.426012438683473e-05,
      "loss": 1.7096,
      "step": 386
    },
    {
      "epoch": 0.025461782653749362,
      "grad_norm": 0.6970518231391907,
      "learning_rate": 8.418205722766374e-05,
      "loss": 1.6868,
      "step": 387
    },
    {
      "epoch": 0.025527575373784892,
      "grad_norm": 0.7553730010986328,
      "learning_rate": 8.410383330766478e-05,
      "loss": 1.7047,
      "step": 388
    },
    {
      "epoch": 0.02559336809382042,
      "grad_norm": 0.7503629326820374,
      "learning_rate": 8.402545298557712e-05,
      "loss": 1.7693,
      "step": 389
    },
    {
      "epoch": 0.025659160813855948,
      "grad_norm": 0.8078016638755798,
      "learning_rate": 8.394691662085731e-05,
      "loss": 1.665,
      "step": 390
    },
    {
      "epoch": 0.025724953533891474,
      "grad_norm": 0.7687540054321289,
      "learning_rate": 8.38682245736775e-05,
      "loss": 1.6099,
      "step": 391
    },
    {
      "epoch": 0.025790746253927004,
      "grad_norm": 0.8783352375030518,
      "learning_rate": 8.378937720492384e-05,
      "loss": 1.8321,
      "step": 392
    },
    {
      "epoch": 0.02585653897396253,
      "grad_norm": 0.7511425614356995,
      "learning_rate": 8.371037487619477e-05,
      "loss": 1.532,
      "step": 393
    },
    {
      "epoch": 0.02592233169399806,
      "grad_norm": 0.8450592756271362,
      "learning_rate": 8.363121794979938e-05,
      "loss": 1.5987,
      "step": 394
    },
    {
      "epoch": 0.025988124414033587,
      "grad_norm": 0.9435873031616211,
      "learning_rate": 8.355190678875578e-05,
      "loss": 1.6183,
      "step": 395
    },
    {
      "epoch": 0.026053917134069116,
      "grad_norm": 0.8024054169654846,
      "learning_rate": 8.347244175678938e-05,
      "loss": 1.6093,
      "step": 396
    },
    {
      "epoch": 0.026119709854104643,
      "grad_norm": 0.8227169513702393,
      "learning_rate": 8.33928232183313e-05,
      "loss": 1.4607,
      "step": 397
    },
    {
      "epoch": 0.026185502574140172,
      "grad_norm": 0.9674575328826904,
      "learning_rate": 8.331305153851658e-05,
      "loss": 1.5625,
      "step": 398
    },
    {
      "epoch": 0.0262512952941757,
      "grad_norm": 0.908593475818634,
      "learning_rate": 8.323312708318262e-05,
      "loss": 1.5447,
      "step": 399
    },
    {
      "epoch": 0.02631708801421123,
      "grad_norm": 1.2496719360351562,
      "learning_rate": 8.315305021886746e-05,
      "loss": 1.3417,
      "step": 400
    },
    {
      "epoch": 0.026382880734246755,
      "grad_norm": 0.5647311806678772,
      "learning_rate": 8.307282131280804e-05,
      "loss": 1.715,
      "step": 401
    },
    {
      "epoch": 0.026448673454282284,
      "grad_norm": 0.5769903063774109,
      "learning_rate": 8.299244073293866e-05,
      "loss": 1.7535,
      "step": 402
    },
    {
      "epoch": 0.02651446617431781,
      "grad_norm": 0.615765392780304,
      "learning_rate": 8.291190884788915e-05,
      "loss": 1.7596,
      "step": 403
    },
    {
      "epoch": 0.02658025889435334,
      "grad_norm": 0.6222158074378967,
      "learning_rate": 8.283122602698323e-05,
      "loss": 1.78,
      "step": 404
    },
    {
      "epoch": 0.026646051614388867,
      "grad_norm": 0.616362452507019,
      "learning_rate": 8.275039264023683e-05,
      "loss": 1.6605,
      "step": 405
    },
    {
      "epoch": 0.026711844334424396,
      "grad_norm": 0.63614422082901,
      "learning_rate": 8.26694090583564e-05,
      "loss": 1.6457,
      "step": 406
    },
    {
      "epoch": 0.026777637054459923,
      "grad_norm": 0.5714118480682373,
      "learning_rate": 8.258827565273718e-05,
      "loss": 1.5967,
      "step": 407
    },
    {
      "epoch": 0.026843429774495452,
      "grad_norm": 0.5620883107185364,
      "learning_rate": 8.250699279546151e-05,
      "loss": 1.5886,
      "step": 408
    },
    {
      "epoch": 0.02690922249453098,
      "grad_norm": 0.593058168888092,
      "learning_rate": 8.242556085929711e-05,
      "loss": 1.7853,
      "step": 409
    },
    {
      "epoch": 0.02697501521456651,
      "grad_norm": 0.6109753251075745,
      "learning_rate": 8.23439802176954e-05,
      "loss": 1.8228,
      "step": 410
    },
    {
      "epoch": 0.027040807934602035,
      "grad_norm": 0.5662338733673096,
      "learning_rate": 8.226225124478979e-05,
      "loss": 1.7658,
      "step": 411
    },
    {
      "epoch": 0.027106600654637564,
      "grad_norm": 0.586213231086731,
      "learning_rate": 8.218037431539391e-05,
      "loss": 1.7126,
      "step": 412
    },
    {
      "epoch": 0.02717239337467309,
      "grad_norm": 0.5898706912994385,
      "learning_rate": 8.209834980499995e-05,
      "loss": 1.6058,
      "step": 413
    },
    {
      "epoch": 0.02723818609470862,
      "grad_norm": 0.6345718502998352,
      "learning_rate": 8.201617808977689e-05,
      "loss": 1.6733,
      "step": 414
    },
    {
      "epoch": 0.02730397881474415,
      "grad_norm": 0.6307375431060791,
      "learning_rate": 8.193385954656883e-05,
      "loss": 1.6946,
      "step": 415
    },
    {
      "epoch": 0.027369771534779676,
      "grad_norm": 0.6170186996459961,
      "learning_rate": 8.185139455289322e-05,
      "loss": 1.6548,
      "step": 416
    },
    {
      "epoch": 0.027435564254815206,
      "grad_norm": 0.6755112409591675,
      "learning_rate": 8.17687834869391e-05,
      "loss": 1.814,
      "step": 417
    },
    {
      "epoch": 0.027501356974850732,
      "grad_norm": 0.636673629283905,
      "learning_rate": 8.16860267275655e-05,
      "loss": 1.7344,
      "step": 418
    },
    {
      "epoch": 0.027567149694886262,
      "grad_norm": 0.6432463526725769,
      "learning_rate": 8.160312465429952e-05,
      "loss": 1.5055,
      "step": 419
    },
    {
      "epoch": 0.02763294241492179,
      "grad_norm": 0.6000040769577026,
      "learning_rate": 8.152007764733471e-05,
      "loss": 1.4705,
      "step": 420
    },
    {
      "epoch": 0.027698735134957318,
      "grad_norm": 0.7080824375152588,
      "learning_rate": 8.14368860875293e-05,
      "loss": 1.7144,
      "step": 421
    },
    {
      "epoch": 0.027764527854992845,
      "grad_norm": 0.6095240712165833,
      "learning_rate": 8.135355035640444e-05,
      "loss": 1.5537,
      "step": 422
    },
    {
      "epoch": 0.027830320575028374,
      "grad_norm": 0.684853196144104,
      "learning_rate": 8.127007083614245e-05,
      "loss": 1.5419,
      "step": 423
    },
    {
      "epoch": 0.0278961132950639,
      "grad_norm": 0.6135299205780029,
      "learning_rate": 8.118644790958509e-05,
      "loss": 1.5624,
      "step": 424
    },
    {
      "epoch": 0.02796190601509943,
      "grad_norm": 0.6529998779296875,
      "learning_rate": 8.110268196023179e-05,
      "loss": 1.6981,
      "step": 425
    },
    {
      "epoch": 0.028027698735134957,
      "grad_norm": 0.7151209115982056,
      "learning_rate": 8.101877337223786e-05,
      "loss": 1.6974,
      "step": 426
    },
    {
      "epoch": 0.028093491455170486,
      "grad_norm": 0.6477805972099304,
      "learning_rate": 8.093472253041281e-05,
      "loss": 1.6776,
      "step": 427
    },
    {
      "epoch": 0.028159284175206013,
      "grad_norm": 0.6722093224525452,
      "learning_rate": 8.085052982021847e-05,
      "loss": 1.6938,
      "step": 428
    },
    {
      "epoch": 0.028225076895241542,
      "grad_norm": 0.6491847038269043,
      "learning_rate": 8.076619562776737e-05,
      "loss": 1.638,
      "step": 429
    },
    {
      "epoch": 0.02829086961527707,
      "grad_norm": 0.6507553458213806,
      "learning_rate": 8.06817203398208e-05,
      "loss": 1.5585,
      "step": 430
    },
    {
      "epoch": 0.0283566623353126,
      "grad_norm": 0.6638416051864624,
      "learning_rate": 8.059710434378715e-05,
      "loss": 1.5053,
      "step": 431
    },
    {
      "epoch": 0.028422455055348125,
      "grad_norm": 0.6749764680862427,
      "learning_rate": 8.051234802772017e-05,
      "loss": 1.6681,
      "step": 432
    },
    {
      "epoch": 0.028488247775383654,
      "grad_norm": 0.6888801455497742,
      "learning_rate": 8.042745178031702e-05,
      "loss": 1.5055,
      "step": 433
    },
    {
      "epoch": 0.02855404049541918,
      "grad_norm": 0.6788802742958069,
      "learning_rate": 8.034241599091665e-05,
      "loss": 1.6874,
      "step": 434
    },
    {
      "epoch": 0.02861983321545471,
      "grad_norm": 0.6617124676704407,
      "learning_rate": 8.025724104949799e-05,
      "loss": 1.5684,
      "step": 435
    },
    {
      "epoch": 0.028685625935490237,
      "grad_norm": 0.6838613748550415,
      "learning_rate": 8.017192734667802e-05,
      "loss": 1.5228,
      "step": 436
    },
    {
      "epoch": 0.028751418655525766,
      "grad_norm": 0.7063144445419312,
      "learning_rate": 8.008647527371023e-05,
      "loss": 1.476,
      "step": 437
    },
    {
      "epoch": 0.028817211375561293,
      "grad_norm": 0.7079107761383057,
      "learning_rate": 8.000088522248255e-05,
      "loss": 1.4906,
      "step": 438
    },
    {
      "epoch": 0.028883004095596822,
      "grad_norm": 0.7689743041992188,
      "learning_rate": 7.991515758551577e-05,
      "loss": 1.5851,
      "step": 439
    },
    {
      "epoch": 0.02894879681563235,
      "grad_norm": 0.807818591594696,
      "learning_rate": 7.982929275596166e-05,
      "loss": 1.6402,
      "step": 440
    },
    {
      "epoch": 0.02901458953566788,
      "grad_norm": 0.7465726733207703,
      "learning_rate": 7.97432911276011e-05,
      "loss": 1.719,
      "step": 441
    },
    {
      "epoch": 0.029080382255703405,
      "grad_norm": 0.7983351945877075,
      "learning_rate": 7.965715309484237e-05,
      "loss": 1.6622,
      "step": 442
    },
    {
      "epoch": 0.029146174975738934,
      "grad_norm": 0.8748970031738281,
      "learning_rate": 7.957087905271934e-05,
      "loss": 1.5992,
      "step": 443
    },
    {
      "epoch": 0.029211967695774464,
      "grad_norm": 0.7884351015090942,
      "learning_rate": 7.948446939688956e-05,
      "loss": 1.4298,
      "step": 444
    },
    {
      "epoch": 0.02927776041580999,
      "grad_norm": 0.7786914110183716,
      "learning_rate": 7.939792452363259e-05,
      "loss": 1.4742,
      "step": 445
    },
    {
      "epoch": 0.02934355313584552,
      "grad_norm": 0.8521863222122192,
      "learning_rate": 7.931124482984802e-05,
      "loss": 1.6724,
      "step": 446
    },
    {
      "epoch": 0.029409345855881047,
      "grad_norm": 0.8581932187080383,
      "learning_rate": 7.92244307130538e-05,
      "loss": 1.4776,
      "step": 447
    },
    {
      "epoch": 0.029475138575916576,
      "grad_norm": 0.8381912112236023,
      "learning_rate": 7.913748257138434e-05,
      "loss": 1.3711,
      "step": 448
    },
    {
      "epoch": 0.029540931295952103,
      "grad_norm": 0.9005985260009766,
      "learning_rate": 7.905040080358868e-05,
      "loss": 1.4701,
      "step": 449
    },
    {
      "epoch": 0.029606724015987632,
      "grad_norm": 1.295373558998108,
      "learning_rate": 7.896318580902867e-05,
      "loss": 1.3209,
      "step": 450
    },
    {
      "epoch": 0.02967251673602316,
      "grad_norm": 0.5462653040885925,
      "learning_rate": 7.887583798767717e-05,
      "loss": 1.716,
      "step": 451
    },
    {
      "epoch": 0.02973830945605869,
      "grad_norm": 0.5886077284812927,
      "learning_rate": 7.878835774011615e-05,
      "loss": 1.7081,
      "step": 452
    },
    {
      "epoch": 0.029804102176094215,
      "grad_norm": 0.5938754081726074,
      "learning_rate": 7.870074546753497e-05,
      "loss": 1.714,
      "step": 453
    },
    {
      "epoch": 0.029869894896129744,
      "grad_norm": 0.6293351054191589,
      "learning_rate": 7.861300157172837e-05,
      "loss": 1.7798,
      "step": 454
    },
    {
      "epoch": 0.02993568761616527,
      "grad_norm": 0.5989140868186951,
      "learning_rate": 7.85251264550948e-05,
      "loss": 1.7753,
      "step": 455
    },
    {
      "epoch": 0.0300014803362008,
      "grad_norm": 0.5683930516242981,
      "learning_rate": 7.843712052063446e-05,
      "loss": 1.5767,
      "step": 456
    },
    {
      "epoch": 0.030067273056236327,
      "grad_norm": 0.6076240539550781,
      "learning_rate": 7.834898417194747e-05,
      "loss": 1.737,
      "step": 457
    },
    {
      "epoch": 0.030133065776271856,
      "grad_norm": 0.6135837435722351,
      "learning_rate": 7.826071781323207e-05,
      "loss": 1.7466,
      "step": 458
    },
    {
      "epoch": 0.030198858496307383,
      "grad_norm": 0.6038995981216431,
      "learning_rate": 7.817232184928276e-05,
      "loss": 1.5442,
      "step": 459
    },
    {
      "epoch": 0.030264651216342912,
      "grad_norm": 0.6276111602783203,
      "learning_rate": 7.808379668548834e-05,
      "loss": 1.6534,
      "step": 460
    },
    {
      "epoch": 0.03033044393637844,
      "grad_norm": 0.6474449634552002,
      "learning_rate": 7.799514272783014e-05,
      "loss": 1.8461,
      "step": 461
    },
    {
      "epoch": 0.03039623665641397,
      "grad_norm": 0.6045409440994263,
      "learning_rate": 7.790636038288023e-05,
      "loss": 1.6837,
      "step": 462
    },
    {
      "epoch": 0.030462029376449495,
      "grad_norm": 0.583416223526001,
      "learning_rate": 7.781745005779938e-05,
      "loss": 1.7384,
      "step": 463
    },
    {
      "epoch": 0.030527822096485024,
      "grad_norm": 0.6061449646949768,
      "learning_rate": 7.772841216033533e-05,
      "loss": 1.6841,
      "step": 464
    },
    {
      "epoch": 0.03059361481652055,
      "grad_norm": 0.6019045114517212,
      "learning_rate": 7.763924709882086e-05,
      "loss": 1.6674,
      "step": 465
    },
    {
      "epoch": 0.03065940753655608,
      "grad_norm": 0.6233644485473633,
      "learning_rate": 7.754995528217194e-05,
      "loss": 1.6055,
      "step": 466
    },
    {
      "epoch": 0.030725200256591607,
      "grad_norm": 0.6329379081726074,
      "learning_rate": 7.746053711988583e-05,
      "loss": 1.7531,
      "step": 467
    },
    {
      "epoch": 0.030790992976627136,
      "grad_norm": 0.6121736168861389,
      "learning_rate": 7.737099302203923e-05,
      "loss": 1.5287,
      "step": 468
    },
    {
      "epoch": 0.030856785696662663,
      "grad_norm": 0.6427604556083679,
      "learning_rate": 7.728132339928638e-05,
      "loss": 1.8374,
      "step": 469
    },
    {
      "epoch": 0.030922578416698192,
      "grad_norm": 0.659582793712616,
      "learning_rate": 7.719152866285721e-05,
      "loss": 1.6771,
      "step": 470
    },
    {
      "epoch": 0.03098837113673372,
      "grad_norm": 0.6290009021759033,
      "learning_rate": 7.710160922455539e-05,
      "loss": 1.7209,
      "step": 471
    },
    {
      "epoch": 0.03105416385676925,
      "grad_norm": 0.6897339224815369,
      "learning_rate": 7.70115654967565e-05,
      "loss": 1.5573,
      "step": 472
    },
    {
      "epoch": 0.031119956576804778,
      "grad_norm": 0.6670756936073303,
      "learning_rate": 7.692139789240611e-05,
      "loss": 1.6864,
      "step": 473
    },
    {
      "epoch": 0.031185749296840305,
      "grad_norm": 0.6782425045967102,
      "learning_rate": 7.68311068250179e-05,
      "loss": 1.7276,
      "step": 474
    },
    {
      "epoch": 0.031251542016875834,
      "grad_norm": 0.6851603388786316,
      "learning_rate": 7.674069270867181e-05,
      "loss": 1.6012,
      "step": 475
    },
    {
      "epoch": 0.03131733473691136,
      "grad_norm": 0.6849592328071594,
      "learning_rate": 7.665015595801197e-05,
      "loss": 1.6589,
      "step": 476
    },
    {
      "epoch": 0.03138312745694689,
      "grad_norm": 0.6817401051521301,
      "learning_rate": 7.6559496988245e-05,
      "loss": 1.5357,
      "step": 477
    },
    {
      "epoch": 0.03144892017698242,
      "grad_norm": 0.6632989645004272,
      "learning_rate": 7.646871621513807e-05,
      "loss": 1.6905,
      "step": 478
    },
    {
      "epoch": 0.031514712897017946,
      "grad_norm": 0.7022443413734436,
      "learning_rate": 7.637781405501681e-05,
      "loss": 1.5377,
      "step": 479
    },
    {
      "epoch": 0.03158050561705347,
      "grad_norm": 0.6468812227249146,
      "learning_rate": 7.628679092476367e-05,
      "loss": 1.6535,
      "step": 480
    },
    {
      "epoch": 0.031646298337089,
      "grad_norm": 0.6711928844451904,
      "learning_rate": 7.619564724181578e-05,
      "loss": 1.5319,
      "step": 481
    },
    {
      "epoch": 0.03171209105712453,
      "grad_norm": 0.8006902933120728,
      "learning_rate": 7.610438342416319e-05,
      "loss": 1.6375,
      "step": 482
    },
    {
      "epoch": 0.03177788377716006,
      "grad_norm": 0.6482203602790833,
      "learning_rate": 7.601299989034689e-05,
      "loss": 1.4045,
      "step": 483
    },
    {
      "epoch": 0.031843676497195585,
      "grad_norm": 0.7363543510437012,
      "learning_rate": 7.592149705945686e-05,
      "loss": 1.8084,
      "step": 484
    },
    {
      "epoch": 0.03190946921723111,
      "grad_norm": 0.7123915553092957,
      "learning_rate": 7.582987535113023e-05,
      "loss": 1.5565,
      "step": 485
    },
    {
      "epoch": 0.031975261937266644,
      "grad_norm": 0.740925133228302,
      "learning_rate": 7.573813518554925e-05,
      "loss": 1.5924,
      "step": 486
    },
    {
      "epoch": 0.03204105465730217,
      "grad_norm": 0.6997798085212708,
      "learning_rate": 7.56462769834395e-05,
      "loss": 1.6812,
      "step": 487
    },
    {
      "epoch": 0.0321068473773377,
      "grad_norm": 0.6777325868606567,
      "learning_rate": 7.555430116606778e-05,
      "loss": 1.5262,
      "step": 488
    },
    {
      "epoch": 0.03217264009737322,
      "grad_norm": 0.7830193638801575,
      "learning_rate": 7.546220815524036e-05,
      "loss": 1.5377,
      "step": 489
    },
    {
      "epoch": 0.032238432817408756,
      "grad_norm": 0.709225594997406,
      "learning_rate": 7.536999837330099e-05,
      "loss": 1.6136,
      "step": 490
    },
    {
      "epoch": 0.03230422553744428,
      "grad_norm": 0.7750362753868103,
      "learning_rate": 7.527767224312883e-05,
      "loss": 1.6665,
      "step": 491
    },
    {
      "epoch": 0.03237001825747981,
      "grad_norm": 0.7182449698448181,
      "learning_rate": 7.51852301881367e-05,
      "loss": 1.6616,
      "step": 492
    },
    {
      "epoch": 0.032435810977515335,
      "grad_norm": 0.7855557799339294,
      "learning_rate": 7.509267263226905e-05,
      "loss": 1.4291,
      "step": 493
    },
    {
      "epoch": 0.03250160369755087,
      "grad_norm": 0.8149724006652832,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.631,
      "step": 494
    },
    {
      "epoch": 0.032567396417586394,
      "grad_norm": 0.8115766644477844,
      "learning_rate": 7.490721271633145e-05,
      "loss": 1.5897,
      "step": 495
    },
    {
      "epoch": 0.03263318913762192,
      "grad_norm": 0.7797816395759583,
      "learning_rate": 7.481431120679106e-05,
      "loss": 1.5576,
      "step": 496
    },
    {
      "epoch": 0.03269898185765745,
      "grad_norm": 0.8388152718544006,
      "learning_rate": 7.472129589743033e-05,
      "loss": 1.5128,
      "step": 497
    },
    {
      "epoch": 0.03276477457769298,
      "grad_norm": 0.8746300935745239,
      "learning_rate": 7.462816721482274e-05,
      "loss": 1.3431,
      "step": 498
    },
    {
      "epoch": 0.032830567297728507,
      "grad_norm": 0.9929096102714539,
      "learning_rate": 7.453492558606157e-05,
      "loss": 1.6894,
      "step": 499
    },
    {
      "epoch": 0.03289636001776403,
      "grad_norm": 1.0941029787063599,
      "learning_rate": 7.44415714387582e-05,
      "loss": 1.122,
      "step": 500
    },
    {
      "epoch": 0.03296215273779956,
      "grad_norm": 0.5158244967460632,
      "learning_rate": 7.434810520103995e-05,
      "loss": 1.6916,
      "step": 501
    },
    {
      "epoch": 0.03302794545783509,
      "grad_norm": 0.5751031637191772,
      "learning_rate": 7.425452730154823e-05,
      "loss": 1.6563,
      "step": 502
    },
    {
      "epoch": 0.03309373817787062,
      "grad_norm": 0.6221696138381958,
      "learning_rate": 7.416083816943653e-05,
      "loss": 1.6673,
      "step": 503
    },
    {
      "epoch": 0.033159530897906145,
      "grad_norm": 0.6054139137268066,
      "learning_rate": 7.406703823436845e-05,
      "loss": 1.7057,
      "step": 504
    },
    {
      "epoch": 0.03322532361794168,
      "grad_norm": 0.6290459036827087,
      "learning_rate": 7.397312792651571e-05,
      "loss": 1.8151,
      "step": 505
    },
    {
      "epoch": 0.033291116337977204,
      "grad_norm": 0.5580724477767944,
      "learning_rate": 7.38791076765563e-05,
      "loss": 1.4864,
      "step": 506
    },
    {
      "epoch": 0.03335690905801273,
      "grad_norm": 0.5801160931587219,
      "learning_rate": 7.378497791567232e-05,
      "loss": 1.7467,
      "step": 507
    },
    {
      "epoch": 0.03342270177804826,
      "grad_norm": 0.5747870206832886,
      "learning_rate": 7.36907390755481e-05,
      "loss": 1.7234,
      "step": 508
    },
    {
      "epoch": 0.03348849449808379,
      "grad_norm": 0.5852952599525452,
      "learning_rate": 7.359639158836828e-05,
      "loss": 1.7107,
      "step": 509
    },
    {
      "epoch": 0.033554287218119316,
      "grad_norm": 0.5938292145729065,
      "learning_rate": 7.350193588681569e-05,
      "loss": 1.7132,
      "step": 510
    },
    {
      "epoch": 0.03362007993815484,
      "grad_norm": 0.5472443103790283,
      "learning_rate": 7.340737240406945e-05,
      "loss": 1.6635,
      "step": 511
    },
    {
      "epoch": 0.03368587265819037,
      "grad_norm": 0.5958325862884521,
      "learning_rate": 7.331270157380303e-05,
      "loss": 1.8009,
      "step": 512
    },
    {
      "epoch": 0.0337516653782259,
      "grad_norm": 0.5720498561859131,
      "learning_rate": 7.321792383018213e-05,
      "loss": 1.6974,
      "step": 513
    },
    {
      "epoch": 0.03381745809826143,
      "grad_norm": 0.5843756198883057,
      "learning_rate": 7.312303960786278e-05,
      "loss": 1.5999,
      "step": 514
    },
    {
      "epoch": 0.033883250818296955,
      "grad_norm": 0.5933427810668945,
      "learning_rate": 7.302804934198936e-05,
      "loss": 1.7271,
      "step": 515
    },
    {
      "epoch": 0.03394904353833248,
      "grad_norm": 0.5814589858055115,
      "learning_rate": 7.293295346819253e-05,
      "loss": 1.5918,
      "step": 516
    },
    {
      "epoch": 0.034014836258368014,
      "grad_norm": 0.6152416467666626,
      "learning_rate": 7.283775242258728e-05,
      "loss": 1.7409,
      "step": 517
    },
    {
      "epoch": 0.03408062897840354,
      "grad_norm": 0.5571946501731873,
      "learning_rate": 7.274244664177097e-05,
      "loss": 1.59,
      "step": 518
    },
    {
      "epoch": 0.03414642169843907,
      "grad_norm": 0.7412462830543518,
      "learning_rate": 7.264703656282122e-05,
      "loss": 1.7138,
      "step": 519
    },
    {
      "epoch": 0.03421221441847459,
      "grad_norm": 0.6679773330688477,
      "learning_rate": 7.2551522623294e-05,
      "loss": 1.7234,
      "step": 520
    },
    {
      "epoch": 0.034278007138510126,
      "grad_norm": 0.6583961844444275,
      "learning_rate": 7.245590526122159e-05,
      "loss": 1.7472,
      "step": 521
    },
    {
      "epoch": 0.03434379985854565,
      "grad_norm": 0.6237342953681946,
      "learning_rate": 7.236018491511053e-05,
      "loss": 1.5379,
      "step": 522
    },
    {
      "epoch": 0.03440959257858118,
      "grad_norm": 0.601783275604248,
      "learning_rate": 7.226436202393973e-05,
      "loss": 1.5308,
      "step": 523
    },
    {
      "epoch": 0.034475385298616705,
      "grad_norm": 0.6741359829902649,
      "learning_rate": 7.216843702715831e-05,
      "loss": 1.65,
      "step": 524
    },
    {
      "epoch": 0.03454117801865224,
      "grad_norm": 0.6531434059143066,
      "learning_rate": 7.207241036468368e-05,
      "loss": 1.577,
      "step": 525
    },
    {
      "epoch": 0.034606970738687765,
      "grad_norm": 0.6689597964286804,
      "learning_rate": 7.197628247689951e-05,
      "loss": 1.6195,
      "step": 526
    },
    {
      "epoch": 0.03467276345872329,
      "grad_norm": 0.6538329124450684,
      "learning_rate": 7.188005380465364e-05,
      "loss": 1.5067,
      "step": 527
    },
    {
      "epoch": 0.03473855617875882,
      "grad_norm": 0.6358965635299683,
      "learning_rate": 7.178372478925617e-05,
      "loss": 1.6048,
      "step": 528
    },
    {
      "epoch": 0.03480434889879435,
      "grad_norm": 0.6213113069534302,
      "learning_rate": 7.168729587247735e-05,
      "loss": 1.5352,
      "step": 529
    },
    {
      "epoch": 0.03487014161882988,
      "grad_norm": 0.656021237373352,
      "learning_rate": 7.159076749654559e-05,
      "loss": 1.6022,
      "step": 530
    },
    {
      "epoch": 0.0349359343388654,
      "grad_norm": 0.7110906839370728,
      "learning_rate": 7.149414010414541e-05,
      "loss": 1.6512,
      "step": 531
    },
    {
      "epoch": 0.035001727058900936,
      "grad_norm": 0.6649793982505798,
      "learning_rate": 7.139741413841548e-05,
      "loss": 1.5512,
      "step": 532
    },
    {
      "epoch": 0.03506751977893646,
      "grad_norm": 0.6741034388542175,
      "learning_rate": 7.130059004294647e-05,
      "loss": 1.5785,
      "step": 533
    },
    {
      "epoch": 0.03513331249897199,
      "grad_norm": 0.7132993340492249,
      "learning_rate": 7.12036682617791e-05,
      "loss": 1.6398,
      "step": 534
    },
    {
      "epoch": 0.035199105219007515,
      "grad_norm": 0.6859109401702881,
      "learning_rate": 7.110664923940209e-05,
      "loss": 1.6834,
      "step": 535
    },
    {
      "epoch": 0.03526489793904305,
      "grad_norm": 0.7714479565620422,
      "learning_rate": 7.10095334207501e-05,
      "loss": 1.537,
      "step": 536
    },
    {
      "epoch": 0.035330690659078574,
      "grad_norm": 0.6838414072990417,
      "learning_rate": 7.091232125120171e-05,
      "loss": 1.4921,
      "step": 537
    },
    {
      "epoch": 0.0353964833791141,
      "grad_norm": 0.7683415412902832,
      "learning_rate": 7.081501317657739e-05,
      "loss": 1.5517,
      "step": 538
    },
    {
      "epoch": 0.03546227609914963,
      "grad_norm": 0.8450960516929626,
      "learning_rate": 7.07176096431374e-05,
      "loss": 1.7357,
      "step": 539
    },
    {
      "epoch": 0.03552806881918516,
      "grad_norm": 0.7302371859550476,
      "learning_rate": 7.06201110975798e-05,
      "loss": 1.5579,
      "step": 540
    },
    {
      "epoch": 0.035593861539220686,
      "grad_norm": 0.7548546195030212,
      "learning_rate": 7.052251798703838e-05,
      "loss": 1.4625,
      "step": 541
    },
    {
      "epoch": 0.03565965425925621,
      "grad_norm": 0.7929666638374329,
      "learning_rate": 7.042483075908062e-05,
      "loss": 1.6235,
      "step": 542
    },
    {
      "epoch": 0.03572544697929174,
      "grad_norm": 0.719232976436615,
      "learning_rate": 7.032704986170559e-05,
      "loss": 1.4643,
      "step": 543
    },
    {
      "epoch": 0.03579123969932727,
      "grad_norm": 0.791790246963501,
      "learning_rate": 7.022917574334199e-05,
      "loss": 1.6799,
      "step": 544
    },
    {
      "epoch": 0.0358570324193628,
      "grad_norm": 0.8136597275733948,
      "learning_rate": 7.013120885284598e-05,
      "loss": 1.5832,
      "step": 545
    },
    {
      "epoch": 0.035922825139398325,
      "grad_norm": 0.8112201690673828,
      "learning_rate": 7.003314963949922e-05,
      "loss": 1.6149,
      "step": 546
    },
    {
      "epoch": 0.03598861785943385,
      "grad_norm": 0.8147010207176208,
      "learning_rate": 6.993499855300673e-05,
      "loss": 1.5676,
      "step": 547
    },
    {
      "epoch": 0.036054410579469384,
      "grad_norm": 1.0767403841018677,
      "learning_rate": 6.983675604349493e-05,
      "loss": 1.742,
      "step": 548
    },
    {
      "epoch": 0.03612020329950491,
      "grad_norm": 0.9000462889671326,
      "learning_rate": 6.973842256150941e-05,
      "loss": 1.3218,
      "step": 549
    },
    {
      "epoch": 0.03618599601954044,
      "grad_norm": 1.4560507535934448,
      "learning_rate": 6.963999855801308e-05,
      "loss": 1.486,
      "step": 550
    },
    {
      "epoch": 0.03625178873957596,
      "grad_norm": 0.5187786221504211,
      "learning_rate": 6.954148448438389e-05,
      "loss": 1.6111,
      "step": 551
    },
    {
      "epoch": 0.036317581459611496,
      "grad_norm": 0.5974017381668091,
      "learning_rate": 6.944288079241293e-05,
      "loss": 1.7928,
      "step": 552
    },
    {
      "epoch": 0.03638337417964702,
      "grad_norm": 0.5858004689216614,
      "learning_rate": 6.934418793430221e-05,
      "loss": 1.6127,
      "step": 553
    },
    {
      "epoch": 0.03644916689968255,
      "grad_norm": 0.5808974504470825,
      "learning_rate": 6.924540636266272e-05,
      "loss": 1.8917,
      "step": 554
    },
    {
      "epoch": 0.036514959619718075,
      "grad_norm": 0.5334272980690002,
      "learning_rate": 6.91465365305123e-05,
      "loss": 1.4775,
      "step": 555
    },
    {
      "epoch": 0.03658075233975361,
      "grad_norm": 0.5895127654075623,
      "learning_rate": 6.90475788912735e-05,
      "loss": 1.7911,
      "step": 556
    },
    {
      "epoch": 0.036646545059789135,
      "grad_norm": 0.5641903281211853,
      "learning_rate": 6.894853389877163e-05,
      "loss": 1.6109,
      "step": 557
    },
    {
      "epoch": 0.03671233777982466,
      "grad_norm": 0.5841579437255859,
      "learning_rate": 6.884940200723255e-05,
      "loss": 1.5897,
      "step": 558
    },
    {
      "epoch": 0.03677813049986019,
      "grad_norm": 0.5516168475151062,
      "learning_rate": 6.875018367128065e-05,
      "loss": 1.5645,
      "step": 559
    },
    {
      "epoch": 0.03684392321989572,
      "grad_norm": 0.5693150758743286,
      "learning_rate": 6.86508793459368e-05,
      "loss": 1.6858,
      "step": 560
    },
    {
      "epoch": 0.03690971593993125,
      "grad_norm": 0.5755842924118042,
      "learning_rate": 6.855148948661616e-05,
      "loss": 1.69,
      "step": 561
    },
    {
      "epoch": 0.03697550865996677,
      "grad_norm": 0.5793744325637817,
      "learning_rate": 6.845201454912621e-05,
      "loss": 1.5982,
      "step": 562
    },
    {
      "epoch": 0.037041301380002306,
      "grad_norm": 0.6278342604637146,
      "learning_rate": 6.835245498966461e-05,
      "loss": 1.6091,
      "step": 563
    },
    {
      "epoch": 0.03710709410003783,
      "grad_norm": 0.5617616772651672,
      "learning_rate": 6.825281126481703e-05,
      "loss": 1.546,
      "step": 564
    },
    {
      "epoch": 0.03717288682007336,
      "grad_norm": 0.5803874135017395,
      "learning_rate": 6.815308383155519e-05,
      "loss": 1.7211,
      "step": 565
    },
    {
      "epoch": 0.037238679540108885,
      "grad_norm": 0.5826717019081116,
      "learning_rate": 6.805327314723468e-05,
      "loss": 1.5747,
      "step": 566
    },
    {
      "epoch": 0.03730447226014442,
      "grad_norm": 0.598685622215271,
      "learning_rate": 6.795337966959289e-05,
      "loss": 1.5133,
      "step": 567
    },
    {
      "epoch": 0.037370264980179944,
      "grad_norm": 0.6577322483062744,
      "learning_rate": 6.785340385674688e-05,
      "loss": 1.63,
      "step": 568
    },
    {
      "epoch": 0.03743605770021547,
      "grad_norm": 0.6314491629600525,
      "learning_rate": 6.775334616719136e-05,
      "loss": 1.647,
      "step": 569
    },
    {
      "epoch": 0.037501850420251,
      "grad_norm": 0.6178677678108215,
      "learning_rate": 6.765320705979646e-05,
      "loss": 1.5232,
      "step": 570
    },
    {
      "epoch": 0.03756764314028653,
      "grad_norm": 0.6231119632720947,
      "learning_rate": 6.755298699380573e-05,
      "loss": 1.6916,
      "step": 571
    },
    {
      "epoch": 0.037633435860322056,
      "grad_norm": 0.7361608147621155,
      "learning_rate": 6.745268642883404e-05,
      "loss": 1.5717,
      "step": 572
    },
    {
      "epoch": 0.03769922858035758,
      "grad_norm": 0.635944128036499,
      "learning_rate": 6.735230582486537e-05,
      "loss": 1.5291,
      "step": 573
    },
    {
      "epoch": 0.03776502130039311,
      "grad_norm": 0.657894492149353,
      "learning_rate": 6.725184564225081e-05,
      "loss": 1.583,
      "step": 574
    },
    {
      "epoch": 0.03783081402042864,
      "grad_norm": 0.6822555661201477,
      "learning_rate": 6.715130634170635e-05,
      "loss": 1.5137,
      "step": 575
    },
    {
      "epoch": 0.03789660674046417,
      "grad_norm": 0.6607404351234436,
      "learning_rate": 6.705068838431086e-05,
      "loss": 1.5549,
      "step": 576
    },
    {
      "epoch": 0.037962399460499695,
      "grad_norm": 0.6624849438667297,
      "learning_rate": 6.694999223150395e-05,
      "loss": 1.5693,
      "step": 577
    },
    {
      "epoch": 0.03802819218053522,
      "grad_norm": 0.7212957143783569,
      "learning_rate": 6.684921834508379e-05,
      "loss": 1.6685,
      "step": 578
    },
    {
      "epoch": 0.038093984900570754,
      "grad_norm": 0.7036564350128174,
      "learning_rate": 6.674836718720506e-05,
      "loss": 1.5121,
      "step": 579
    },
    {
      "epoch": 0.03815977762060628,
      "grad_norm": 0.6782734990119934,
      "learning_rate": 6.664743922037684e-05,
      "loss": 1.479,
      "step": 580
    },
    {
      "epoch": 0.03822557034064181,
      "grad_norm": 0.6969690918922424,
      "learning_rate": 6.654643490746042e-05,
      "loss": 1.6161,
      "step": 581
    },
    {
      "epoch": 0.03829136306067733,
      "grad_norm": 0.7635255455970764,
      "learning_rate": 6.644535471166724e-05,
      "loss": 1.7393,
      "step": 582
    },
    {
      "epoch": 0.038357155780712866,
      "grad_norm": 0.7802557945251465,
      "learning_rate": 6.634419909655672e-05,
      "loss": 1.6247,
      "step": 583
    },
    {
      "epoch": 0.03842294850074839,
      "grad_norm": 0.7853055000305176,
      "learning_rate": 6.624296852603419e-05,
      "loss": 1.8486,
      "step": 584
    },
    {
      "epoch": 0.03848874122078392,
      "grad_norm": 0.699921727180481,
      "learning_rate": 6.61416634643487e-05,
      "loss": 1.5465,
      "step": 585
    },
    {
      "epoch": 0.038554533940819445,
      "grad_norm": 0.7867047190666199,
      "learning_rate": 6.604028437609095e-05,
      "loss": 1.6135,
      "step": 586
    },
    {
      "epoch": 0.03862032666085498,
      "grad_norm": 0.816728949546814,
      "learning_rate": 6.593883172619111e-05,
      "loss": 1.7464,
      "step": 587
    },
    {
      "epoch": 0.038686119380890505,
      "grad_norm": 0.7413901686668396,
      "learning_rate": 6.583730597991671e-05,
      "loss": 1.6291,
      "step": 588
    },
    {
      "epoch": 0.03875191210092603,
      "grad_norm": 0.8019276261329651,
      "learning_rate": 6.573570760287048e-05,
      "loss": 1.5239,
      "step": 589
    },
    {
      "epoch": 0.03881770482096156,
      "grad_norm": 0.7708191275596619,
      "learning_rate": 6.563403706098833e-05,
      "loss": 1.6306,
      "step": 590
    },
    {
      "epoch": 0.03888349754099709,
      "grad_norm": 0.7502703070640564,
      "learning_rate": 6.553229482053698e-05,
      "loss": 1.6766,
      "step": 591
    },
    {
      "epoch": 0.03894929026103262,
      "grad_norm": 0.8003950119018555,
      "learning_rate": 6.543048134811209e-05,
      "loss": 1.5102,
      "step": 592
    },
    {
      "epoch": 0.03901508298106814,
      "grad_norm": 0.8051195740699768,
      "learning_rate": 6.532859711063594e-05,
      "loss": 1.4946,
      "step": 593
    },
    {
      "epoch": 0.039080875701103676,
      "grad_norm": 0.7975060939788818,
      "learning_rate": 6.522664257535533e-05,
      "loss": 1.3503,
      "step": 594
    },
    {
      "epoch": 0.0391466684211392,
      "grad_norm": 0.8413264751434326,
      "learning_rate": 6.512461820983946e-05,
      "loss": 1.4129,
      "step": 595
    },
    {
      "epoch": 0.03921246114117473,
      "grad_norm": 0.94151771068573,
      "learning_rate": 6.502252448197782e-05,
      "loss": 1.5474,
      "step": 596
    },
    {
      "epoch": 0.039278253861210255,
      "grad_norm": 0.9534524083137512,
      "learning_rate": 6.49203618599779e-05,
      "loss": 1.7038,
      "step": 597
    },
    {
      "epoch": 0.03934404658124579,
      "grad_norm": 0.959064781665802,
      "learning_rate": 6.481813081236328e-05,
      "loss": 1.5944,
      "step": 598
    },
    {
      "epoch": 0.039409839301281314,
      "grad_norm": 1.0186463594436646,
      "learning_rate": 6.471583180797121e-05,
      "loss": 1.4035,
      "step": 599
    },
    {
      "epoch": 0.03947563202131684,
      "grad_norm": 1.1875278949737549,
      "learning_rate": 6.461346531595065e-05,
      "loss": 1.4765,
      "step": 600
    },
    {
      "epoch": 0.03954142474135237,
      "grad_norm": 0.5883246660232544,
      "learning_rate": 6.45110318057601e-05,
      "loss": 1.8354,
      "step": 601
    },
    {
      "epoch": 0.0396072174613879,
      "grad_norm": 0.6155911684036255,
      "learning_rate": 6.440853174716534e-05,
      "loss": 1.5164,
      "step": 602
    },
    {
      "epoch": 0.039673010181423427,
      "grad_norm": 0.5574658513069153,
      "learning_rate": 6.43059656102374e-05,
      "loss": 1.6834,
      "step": 603
    },
    {
      "epoch": 0.03973880290145895,
      "grad_norm": 0.6176509261131287,
      "learning_rate": 6.420333386535032e-05,
      "loss": 1.8324,
      "step": 604
    },
    {
      "epoch": 0.03980459562149448,
      "grad_norm": 0.545872151851654,
      "learning_rate": 6.410063698317901e-05,
      "loss": 1.6041,
      "step": 605
    },
    {
      "epoch": 0.03987038834153001,
      "grad_norm": 0.60647052526474,
      "learning_rate": 6.399787543469715e-05,
      "loss": 1.5879,
      "step": 606
    },
    {
      "epoch": 0.03993618106156554,
      "grad_norm": 0.566210150718689,
      "learning_rate": 6.389504969117492e-05,
      "loss": 1.7142,
      "step": 607
    },
    {
      "epoch": 0.040001973781601065,
      "grad_norm": 0.6033192276954651,
      "learning_rate": 6.379216022417696e-05,
      "loss": 1.7099,
      "step": 608
    },
    {
      "epoch": 0.04006776650163659,
      "grad_norm": 0.5859748721122742,
      "learning_rate": 6.368920750556012e-05,
      "loss": 1.7776,
      "step": 609
    },
    {
      "epoch": 0.040133559221672124,
      "grad_norm": 0.6001906394958496,
      "learning_rate": 6.358619200747132e-05,
      "loss": 1.8193,
      "step": 610
    },
    {
      "epoch": 0.04019935194170765,
      "grad_norm": 0.5797659754753113,
      "learning_rate": 6.348311420234542e-05,
      "loss": 1.5826,
      "step": 611
    },
    {
      "epoch": 0.04026514466174318,
      "grad_norm": 0.5737758874893188,
      "learning_rate": 6.3379974562903e-05,
      "loss": 1.7274,
      "step": 612
    },
    {
      "epoch": 0.0403309373817787,
      "grad_norm": 0.5750417709350586,
      "learning_rate": 6.32767735621482e-05,
      "loss": 1.7693,
      "step": 613
    },
    {
      "epoch": 0.040396730101814236,
      "grad_norm": 0.5750201940536499,
      "learning_rate": 6.31735116733666e-05,
      "loss": 1.6943,
      "step": 614
    },
    {
      "epoch": 0.04046252282184976,
      "grad_norm": 0.6071341633796692,
      "learning_rate": 6.307018937012303e-05,
      "loss": 1.6225,
      "step": 615
    },
    {
      "epoch": 0.04052831554188529,
      "grad_norm": 0.6592499613761902,
      "learning_rate": 6.296680712625932e-05,
      "loss": 1.6679,
      "step": 616
    },
    {
      "epoch": 0.040594108261920815,
      "grad_norm": 0.5857351422309875,
      "learning_rate": 6.286336541589224e-05,
      "loss": 1.5893,
      "step": 617
    },
    {
      "epoch": 0.04065990098195635,
      "grad_norm": 0.606536328792572,
      "learning_rate": 6.275986471341125e-05,
      "loss": 1.7409,
      "step": 618
    },
    {
      "epoch": 0.040725693701991875,
      "grad_norm": 0.5929151773452759,
      "learning_rate": 6.26563054934764e-05,
      "loss": 1.6339,
      "step": 619
    },
    {
      "epoch": 0.0407914864220274,
      "grad_norm": 0.6137556433677673,
      "learning_rate": 6.255268823101605e-05,
      "loss": 1.6275,
      "step": 620
    },
    {
      "epoch": 0.040857279142062934,
      "grad_norm": 0.6244606971740723,
      "learning_rate": 6.244901340122475e-05,
      "loss": 1.7192,
      "step": 621
    },
    {
      "epoch": 0.04092307186209846,
      "grad_norm": 0.6325397491455078,
      "learning_rate": 6.234528147956108e-05,
      "loss": 1.6312,
      "step": 622
    },
    {
      "epoch": 0.04098886458213399,
      "grad_norm": 0.6136852502822876,
      "learning_rate": 6.224149294174548e-05,
      "loss": 1.564,
      "step": 623
    },
    {
      "epoch": 0.04105465730216951,
      "grad_norm": 0.6611918210983276,
      "learning_rate": 6.213764826375795e-05,
      "loss": 1.5314,
      "step": 624
    },
    {
      "epoch": 0.041120450022205046,
      "grad_norm": 0.6270449757575989,
      "learning_rate": 6.203374792183599e-05,
      "loss": 1.5965,
      "step": 625
    },
    {
      "epoch": 0.04118624274224057,
      "grad_norm": 0.7064681053161621,
      "learning_rate": 6.192979239247243e-05,
      "loss": 1.8145,
      "step": 626
    },
    {
      "epoch": 0.0412520354622761,
      "grad_norm": 0.6724807024002075,
      "learning_rate": 6.182578215241311e-05,
      "loss": 1.6588,
      "step": 627
    },
    {
      "epoch": 0.041317828182311625,
      "grad_norm": 0.6827772259712219,
      "learning_rate": 6.172171767865483e-05,
      "loss": 1.5998,
      "step": 628
    },
    {
      "epoch": 0.04138362090234716,
      "grad_norm": 0.6520280838012695,
      "learning_rate": 6.161759944844308e-05,
      "loss": 1.5482,
      "step": 629
    },
    {
      "epoch": 0.041449413622382685,
      "grad_norm": 0.6610122323036194,
      "learning_rate": 6.15134279392699e-05,
      "loss": 1.5476,
      "step": 630
    },
    {
      "epoch": 0.04151520634241821,
      "grad_norm": 0.6820929646492004,
      "learning_rate": 6.140920362887167e-05,
      "loss": 1.5707,
      "step": 631
    },
    {
      "epoch": 0.04158099906245374,
      "grad_norm": 0.7021801471710205,
      "learning_rate": 6.13049269952269e-05,
      "loss": 1.711,
      "step": 632
    },
    {
      "epoch": 0.04164679178248927,
      "grad_norm": 0.6803610920906067,
      "learning_rate": 6.120059851655408e-05,
      "loss": 1.6353,
      "step": 633
    },
    {
      "epoch": 0.0417125845025248,
      "grad_norm": 0.6335561275482178,
      "learning_rate": 6.109621867130944e-05,
      "loss": 1.4506,
      "step": 634
    },
    {
      "epoch": 0.04177837722256032,
      "grad_norm": 0.679242730140686,
      "learning_rate": 6.0991787938184784e-05,
      "loss": 1.5531,
      "step": 635
    },
    {
      "epoch": 0.04184416994259585,
      "grad_norm": 0.7108798027038574,
      "learning_rate": 6.088730679610535e-05,
      "loss": 1.5429,
      "step": 636
    },
    {
      "epoch": 0.04190996266263138,
      "grad_norm": 0.6999435424804688,
      "learning_rate": 6.0782775724227436e-05,
      "loss": 1.762,
      "step": 637
    },
    {
      "epoch": 0.04197575538266691,
      "grad_norm": 0.7404870390892029,
      "learning_rate": 6.067819520193645e-05,
      "loss": 1.5536,
      "step": 638
    },
    {
      "epoch": 0.042041548102702435,
      "grad_norm": 0.7356504797935486,
      "learning_rate": 6.057356570884449e-05,
      "loss": 1.5618,
      "step": 639
    },
    {
      "epoch": 0.04210734082273796,
      "grad_norm": 0.7335320115089417,
      "learning_rate": 6.046888772478828e-05,
      "loss": 1.557,
      "step": 640
    },
    {
      "epoch": 0.042173133542773494,
      "grad_norm": 0.7280996441841125,
      "learning_rate": 6.0364161729826905e-05,
      "loss": 1.6124,
      "step": 641
    },
    {
      "epoch": 0.04223892626280902,
      "grad_norm": 0.8570873737335205,
      "learning_rate": 6.025938820423964e-05,
      "loss": 1.6069,
      "step": 642
    },
    {
      "epoch": 0.04230471898284455,
      "grad_norm": 0.7949631810188293,
      "learning_rate": 6.015456762852374e-05,
      "loss": 1.5835,
      "step": 643
    },
    {
      "epoch": 0.04237051170288007,
      "grad_norm": 0.8168470859527588,
      "learning_rate": 6.004970048339226e-05,
      "loss": 1.4906,
      "step": 644
    },
    {
      "epoch": 0.042436304422915606,
      "grad_norm": 0.7664223909378052,
      "learning_rate": 5.994478724977175e-05,
      "loss": 1.3437,
      "step": 645
    },
    {
      "epoch": 0.04250209714295113,
      "grad_norm": 0.7634305357933044,
      "learning_rate": 5.9839828408800204e-05,
      "loss": 1.4158,
      "step": 646
    },
    {
      "epoch": 0.04256788986298666,
      "grad_norm": 0.870861291885376,
      "learning_rate": 5.973482444182475e-05,
      "loss": 1.5231,
      "step": 647
    },
    {
      "epoch": 0.042633682583022185,
      "grad_norm": 0.8543813228607178,
      "learning_rate": 5.9629775830399424e-05,
      "loss": 1.3723,
      "step": 648
    },
    {
      "epoch": 0.04269947530305772,
      "grad_norm": 0.9646958708763123,
      "learning_rate": 5.952468305628307e-05,
      "loss": 1.457,
      "step": 649
    },
    {
      "epoch": 0.042765268023093245,
      "grad_norm": 1.210070252418518,
      "learning_rate": 5.941954660143703e-05,
      "loss": 1.4912,
      "step": 650
    },
    {
      "epoch": 0.04283106074312877,
      "grad_norm": 0.5520440936088562,
      "learning_rate": 5.931436694802295e-05,
      "loss": 1.8551,
      "step": 651
    },
    {
      "epoch": 0.042896853463164304,
      "grad_norm": 0.5633511543273926,
      "learning_rate": 5.920914457840063e-05,
      "loss": 1.696,
      "step": 652
    },
    {
      "epoch": 0.04296264618319983,
      "grad_norm": 0.5891466736793518,
      "learning_rate": 5.910387997512573e-05,
      "loss": 1.6773,
      "step": 653
    },
    {
      "epoch": 0.04302843890323536,
      "grad_norm": 0.5962758660316467,
      "learning_rate": 5.899857362094763e-05,
      "loss": 1.6356,
      "step": 654
    },
    {
      "epoch": 0.04309423162327088,
      "grad_norm": 0.6261197328567505,
      "learning_rate": 5.8893225998807166e-05,
      "loss": 1.6317,
      "step": 655
    },
    {
      "epoch": 0.043160024343306416,
      "grad_norm": 0.605209469795227,
      "learning_rate": 5.878783759183442e-05,
      "loss": 1.6793,
      "step": 656
    },
    {
      "epoch": 0.04322581706334194,
      "grad_norm": 0.673530638217926,
      "learning_rate": 5.868240888334653e-05,
      "loss": 1.8195,
      "step": 657
    },
    {
      "epoch": 0.04329160978337747,
      "grad_norm": 0.595020055770874,
      "learning_rate": 5.857694035684545e-05,
      "loss": 1.6684,
      "step": 658
    },
    {
      "epoch": 0.043357402503412995,
      "grad_norm": 0.6050863265991211,
      "learning_rate": 5.847143249601574e-05,
      "loss": 1.6713,
      "step": 659
    },
    {
      "epoch": 0.04342319522344853,
      "grad_norm": 0.6145523190498352,
      "learning_rate": 5.8365885784722394e-05,
      "loss": 1.7704,
      "step": 660
    },
    {
      "epoch": 0.043488987943484055,
      "grad_norm": 0.5903642177581787,
      "learning_rate": 5.826030070700849e-05,
      "loss": 1.8083,
      "step": 661
    },
    {
      "epoch": 0.04355478066351958,
      "grad_norm": 0.6061384081840515,
      "learning_rate": 5.8154677747093134e-05,
      "loss": 1.713,
      "step": 662
    },
    {
      "epoch": 0.04362057338355511,
      "grad_norm": 0.6104039549827576,
      "learning_rate": 5.804901738936914e-05,
      "loss": 1.6264,
      "step": 663
    },
    {
      "epoch": 0.04368636610359064,
      "grad_norm": 0.6567908525466919,
      "learning_rate": 5.794332011840079e-05,
      "loss": 1.6822,
      "step": 664
    },
    {
      "epoch": 0.04375215882362617,
      "grad_norm": 0.6191847324371338,
      "learning_rate": 5.783758641892172e-05,
      "loss": 1.6488,
      "step": 665
    },
    {
      "epoch": 0.04381795154366169,
      "grad_norm": 0.648104727268219,
      "learning_rate": 5.773181677583257e-05,
      "loss": 1.6951,
      "step": 666
    },
    {
      "epoch": 0.04388374426369722,
      "grad_norm": 0.610373854637146,
      "learning_rate": 5.762601167419885e-05,
      "loss": 1.6071,
      "step": 667
    },
    {
      "epoch": 0.04394953698373275,
      "grad_norm": 0.6029400825500488,
      "learning_rate": 5.7520171599248704e-05,
      "loss": 1.5817,
      "step": 668
    },
    {
      "epoch": 0.04401532970376828,
      "grad_norm": 0.5969151854515076,
      "learning_rate": 5.74142970363706e-05,
      "loss": 1.478,
      "step": 669
    },
    {
      "epoch": 0.044081122423803805,
      "grad_norm": 0.6027969717979431,
      "learning_rate": 5.730838847111123e-05,
      "loss": 1.5602,
      "step": 670
    },
    {
      "epoch": 0.04414691514383933,
      "grad_norm": 0.600036084651947,
      "learning_rate": 5.7202446389173223e-05,
      "loss": 1.6526,
      "step": 671
    },
    {
      "epoch": 0.044212707863874864,
      "grad_norm": 0.648833692073822,
      "learning_rate": 5.709647127641286e-05,
      "loss": 1.6957,
      "step": 672
    },
    {
      "epoch": 0.04427850058391039,
      "grad_norm": 0.5996218323707581,
      "learning_rate": 5.6990463618837977e-05,
      "loss": 1.513,
      "step": 673
    },
    {
      "epoch": 0.04434429330394592,
      "grad_norm": 0.6365662217140198,
      "learning_rate": 5.688442390260559e-05,
      "loss": 1.5507,
      "step": 674
    },
    {
      "epoch": 0.04441008602398144,
      "grad_norm": 0.6217913627624512,
      "learning_rate": 5.6778352614019795e-05,
      "loss": 1.6172,
      "step": 675
    },
    {
      "epoch": 0.044475878744016976,
      "grad_norm": 0.6233986020088196,
      "learning_rate": 5.6672250239529465e-05,
      "loss": 1.3737,
      "step": 676
    },
    {
      "epoch": 0.0445416714640525,
      "grad_norm": 0.6476162075996399,
      "learning_rate": 5.6566117265726006e-05,
      "loss": 1.5756,
      "step": 677
    },
    {
      "epoch": 0.04460746418408803,
      "grad_norm": 0.6581618189811707,
      "learning_rate": 5.645995417934119e-05,
      "loss": 1.5962,
      "step": 678
    },
    {
      "epoch": 0.044673256904123555,
      "grad_norm": 0.6630986928939819,
      "learning_rate": 5.635376146724489e-05,
      "loss": 1.6214,
      "step": 679
    },
    {
      "epoch": 0.04473904962415909,
      "grad_norm": 0.7406059503555298,
      "learning_rate": 5.624753961644281e-05,
      "loss": 1.6885,
      "step": 680
    },
    {
      "epoch": 0.044804842344194615,
      "grad_norm": 0.6429552435874939,
      "learning_rate": 5.614128911407431e-05,
      "loss": 1.4918,
      "step": 681
    },
    {
      "epoch": 0.04487063506423014,
      "grad_norm": 0.7244037389755249,
      "learning_rate": 5.603501044741013e-05,
      "loss": 1.5453,
      "step": 682
    },
    {
      "epoch": 0.044936427784265674,
      "grad_norm": 0.7735380530357361,
      "learning_rate": 5.5928704103850206e-05,
      "loss": 1.5844,
      "step": 683
    },
    {
      "epoch": 0.0450022205043012,
      "grad_norm": 0.701229989528656,
      "learning_rate": 5.582237057092137e-05,
      "loss": 1.551,
      "step": 684
    },
    {
      "epoch": 0.04506801322433673,
      "grad_norm": 0.7064613103866577,
      "learning_rate": 5.571601033627514e-05,
      "loss": 1.6503,
      "step": 685
    },
    {
      "epoch": 0.04513380594437225,
      "grad_norm": 0.7237836718559265,
      "learning_rate": 5.5609623887685535e-05,
      "loss": 1.668,
      "step": 686
    },
    {
      "epoch": 0.045199598664407786,
      "grad_norm": 0.7475508451461792,
      "learning_rate": 5.550321171304675e-05,
      "loss": 1.6663,
      "step": 687
    },
    {
      "epoch": 0.04526539138444331,
      "grad_norm": 0.7455587983131409,
      "learning_rate": 5.539677430037098e-05,
      "loss": 1.6195,
      "step": 688
    },
    {
      "epoch": 0.04533118410447884,
      "grad_norm": 0.6819461584091187,
      "learning_rate": 5.5290312137786146e-05,
      "loss": 1.4005,
      "step": 689
    },
    {
      "epoch": 0.045396976824514365,
      "grad_norm": 0.7876448631286621,
      "learning_rate": 5.518382571353369e-05,
      "loss": 1.6189,
      "step": 690
    },
    {
      "epoch": 0.0454627695445499,
      "grad_norm": 0.7936854958534241,
      "learning_rate": 5.507731551596632e-05,
      "loss": 1.7235,
      "step": 691
    },
    {
      "epoch": 0.045528562264585425,
      "grad_norm": 0.7383480668067932,
      "learning_rate": 5.4970782033545774e-05,
      "loss": 1.5017,
      "step": 692
    },
    {
      "epoch": 0.04559435498462095,
      "grad_norm": 0.7927383184432983,
      "learning_rate": 5.486422575484054e-05,
      "loss": 1.5818,
      "step": 693
    },
    {
      "epoch": 0.04566014770465648,
      "grad_norm": 0.8128315806388855,
      "learning_rate": 5.47576471685237e-05,
      "loss": 1.4506,
      "step": 694
    },
    {
      "epoch": 0.04572594042469201,
      "grad_norm": 0.7387059330940247,
      "learning_rate": 5.465104676337062e-05,
      "loss": 1.3782,
      "step": 695
    },
    {
      "epoch": 0.04579173314472754,
      "grad_norm": 0.974277913570404,
      "learning_rate": 5.4544425028256695e-05,
      "loss": 1.681,
      "step": 696
    },
    {
      "epoch": 0.04585752586476306,
      "grad_norm": 0.9682861566543579,
      "learning_rate": 5.443778245215519e-05,
      "loss": 1.6909,
      "step": 697
    },
    {
      "epoch": 0.04592331858479859,
      "grad_norm": 0.8716404438018799,
      "learning_rate": 5.433111952413495e-05,
      "loss": 1.4352,
      "step": 698
    },
    {
      "epoch": 0.04598911130483412,
      "grad_norm": 0.9771693348884583,
      "learning_rate": 5.42244367333581e-05,
      "loss": 1.4673,
      "step": 699
    },
    {
      "epoch": 0.04605490402486965,
      "grad_norm": 1.2287424802780151,
      "learning_rate": 5.411773456907792e-05,
      "loss": 1.2711,
      "step": 700
    },
    {
      "epoch": 0.046120696744905175,
      "grad_norm": 0.5442025065422058,
      "learning_rate": 5.401101352063647e-05,
      "loss": 1.7773,
      "step": 701
    },
    {
      "epoch": 0.0461864894649407,
      "grad_norm": 0.577033519744873,
      "learning_rate": 5.390427407746248e-05,
      "loss": 1.6329,
      "step": 702
    },
    {
      "epoch": 0.046252282184976234,
      "grad_norm": 0.5880780816078186,
      "learning_rate": 5.379751672906902e-05,
      "loss": 1.6615,
      "step": 703
    },
    {
      "epoch": 0.04631807490501176,
      "grad_norm": 0.5862165689468384,
      "learning_rate": 5.369074196505125e-05,
      "loss": 1.7882,
      "step": 704
    },
    {
      "epoch": 0.04638386762504729,
      "grad_norm": 0.5397851467132568,
      "learning_rate": 5.3583950275084206e-05,
      "loss": 1.5977,
      "step": 705
    },
    {
      "epoch": 0.04644966034508281,
      "grad_norm": 0.581055760383606,
      "learning_rate": 5.347714214892058e-05,
      "loss": 1.5843,
      "step": 706
    },
    {
      "epoch": 0.046515453065118346,
      "grad_norm": 0.6341369152069092,
      "learning_rate": 5.33703180763884e-05,
      "loss": 1.698,
      "step": 707
    },
    {
      "epoch": 0.04658124578515387,
      "grad_norm": 0.562896192073822,
      "learning_rate": 5.3263478547388865e-05,
      "loss": 1.6773,
      "step": 708
    },
    {
      "epoch": 0.0466470385051894,
      "grad_norm": 0.5934199094772339,
      "learning_rate": 5.3156624051894e-05,
      "loss": 1.7178,
      "step": 709
    },
    {
      "epoch": 0.04671283122522493,
      "grad_norm": 0.5896904468536377,
      "learning_rate": 5.3049755079944527e-05,
      "loss": 1.5911,
      "step": 710
    },
    {
      "epoch": 0.04677862394526046,
      "grad_norm": 0.6300824284553528,
      "learning_rate": 5.2942872121647546e-05,
      "loss": 1.6214,
      "step": 711
    },
    {
      "epoch": 0.046844416665295985,
      "grad_norm": 0.6136653423309326,
      "learning_rate": 5.2835975667174254e-05,
      "loss": 1.7043,
      "step": 712
    },
    {
      "epoch": 0.04691020938533151,
      "grad_norm": 0.5764832496643066,
      "learning_rate": 5.272906620675779e-05,
      "loss": 1.5961,
      "step": 713
    },
    {
      "epoch": 0.046976002105367044,
      "grad_norm": 0.6364599466323853,
      "learning_rate": 5.2622144230690954e-05,
      "loss": 1.8236,
      "step": 714
    },
    {
      "epoch": 0.04704179482540257,
      "grad_norm": 0.6356618404388428,
      "learning_rate": 5.2515210229323866e-05,
      "loss": 1.6795,
      "step": 715
    },
    {
      "epoch": 0.0471075875454381,
      "grad_norm": 0.6253381371498108,
      "learning_rate": 5.240826469306187e-05,
      "loss": 1.5472,
      "step": 716
    },
    {
      "epoch": 0.04717338026547362,
      "grad_norm": 0.5967807173728943,
      "learning_rate": 5.230130811236316e-05,
      "loss": 1.7268,
      "step": 717
    },
    {
      "epoch": 0.047239172985509156,
      "grad_norm": 0.5806270241737366,
      "learning_rate": 5.2194340977736635e-05,
      "loss": 1.571,
      "step": 718
    },
    {
      "epoch": 0.04730496570554468,
      "grad_norm": 0.6197123527526855,
      "learning_rate": 5.208736377973954e-05,
      "loss": 1.6507,
      "step": 719
    },
    {
      "epoch": 0.04737075842558021,
      "grad_norm": 0.6148327589035034,
      "learning_rate": 5.19803770089753e-05,
      "loss": 1.8198,
      "step": 720
    },
    {
      "epoch": 0.047436551145615735,
      "grad_norm": 0.6454043984413147,
      "learning_rate": 5.187338115609123e-05,
      "loss": 1.4741,
      "step": 721
    },
    {
      "epoch": 0.04750234386565127,
      "grad_norm": 0.6259646415710449,
      "learning_rate": 5.176637671177631e-05,
      "loss": 1.6452,
      "step": 722
    },
    {
      "epoch": 0.047568136585686795,
      "grad_norm": 0.6418277025222778,
      "learning_rate": 5.1659364166758904e-05,
      "loss": 1.4635,
      "step": 723
    },
    {
      "epoch": 0.04763392930572232,
      "grad_norm": 0.6576802134513855,
      "learning_rate": 5.1552344011804554e-05,
      "loss": 1.5849,
      "step": 724
    },
    {
      "epoch": 0.04769972202575785,
      "grad_norm": 0.679136335849762,
      "learning_rate": 5.144531673771363e-05,
      "loss": 1.7297,
      "step": 725
    },
    {
      "epoch": 0.04776551474579338,
      "grad_norm": 0.6577818393707275,
      "learning_rate": 5.133828283531926e-05,
      "loss": 1.7771,
      "step": 726
    },
    {
      "epoch": 0.04783130746582891,
      "grad_norm": 0.6642506718635559,
      "learning_rate": 5.12312427954849e-05,
      "loss": 1.6831,
      "step": 727
    },
    {
      "epoch": 0.04789710018586443,
      "grad_norm": 0.6602113842964172,
      "learning_rate": 5.112419710910213e-05,
      "loss": 1.6407,
      "step": 728
    },
    {
      "epoch": 0.04796289290589996,
      "grad_norm": 0.6676844954490662,
      "learning_rate": 5.101714626708849e-05,
      "loss": 1.5256,
      "step": 729
    },
    {
      "epoch": 0.04802868562593549,
      "grad_norm": 0.6657158136367798,
      "learning_rate": 5.091009076038514e-05,
      "loss": 1.6361,
      "step": 730
    },
    {
      "epoch": 0.04809447834597102,
      "grad_norm": 0.6682079434394836,
      "learning_rate": 5.080303107995461e-05,
      "loss": 1.4751,
      "step": 731
    },
    {
      "epoch": 0.048160271066006545,
      "grad_norm": 0.68694007396698,
      "learning_rate": 5.0695967716778606e-05,
      "loss": 1.6417,
      "step": 732
    },
    {
      "epoch": 0.04822606378604207,
      "grad_norm": 0.726314127445221,
      "learning_rate": 5.058890116185566e-05,
      "loss": 1.5361,
      "step": 733
    },
    {
      "epoch": 0.048291856506077604,
      "grad_norm": 0.6680204272270203,
      "learning_rate": 5.048183190619904e-05,
      "loss": 1.4233,
      "step": 734
    },
    {
      "epoch": 0.04835764922611313,
      "grad_norm": 0.686355710029602,
      "learning_rate": 5.0374760440834335e-05,
      "loss": 1.4506,
      "step": 735
    },
    {
      "epoch": 0.04842344194614866,
      "grad_norm": 0.7238395810127258,
      "learning_rate": 5.026768725679726e-05,
      "loss": 1.5804,
      "step": 736
    },
    {
      "epoch": 0.04848923466618418,
      "grad_norm": 0.7959031462669373,
      "learning_rate": 5.0160612845131414e-05,
      "loss": 1.761,
      "step": 737
    },
    {
      "epoch": 0.04855502738621972,
      "grad_norm": 0.7459606528282166,
      "learning_rate": 5.005353769688611e-05,
      "loss": 1.4302,
      "step": 738
    },
    {
      "epoch": 0.04862082010625524,
      "grad_norm": 0.7083699703216553,
      "learning_rate": 4.994646230311391e-05,
      "loss": 1.5262,
      "step": 739
    },
    {
      "epoch": 0.04868661282629077,
      "grad_norm": 0.7352794408798218,
      "learning_rate": 4.9839387154868584e-05,
      "loss": 1.5299,
      "step": 740
    },
    {
      "epoch": 0.0487524055463263,
      "grad_norm": 0.7063877582550049,
      "learning_rate": 4.973231274320276e-05,
      "loss": 1.6491,
      "step": 741
    },
    {
      "epoch": 0.04881819826636183,
      "grad_norm": 0.7635167837142944,
      "learning_rate": 4.962523955916569e-05,
      "loss": 1.677,
      "step": 742
    },
    {
      "epoch": 0.048883990986397355,
      "grad_norm": 0.756692111492157,
      "learning_rate": 4.951816809380097e-05,
      "loss": 1.4049,
      "step": 743
    },
    {
      "epoch": 0.04894978370643288,
      "grad_norm": 0.7604605555534363,
      "learning_rate": 4.9411098838144346e-05,
      "loss": 1.6448,
      "step": 744
    },
    {
      "epoch": 0.049015576426468414,
      "grad_norm": 0.8182808756828308,
      "learning_rate": 4.9304032283221405e-05,
      "loss": 1.5077,
      "step": 745
    },
    {
      "epoch": 0.04908136914650394,
      "grad_norm": 0.7334157824516296,
      "learning_rate": 4.919696892004539e-05,
      "loss": 1.4763,
      "step": 746
    },
    {
      "epoch": 0.04914716186653947,
      "grad_norm": 0.8388804197311401,
      "learning_rate": 4.908990923961488e-05,
      "loss": 1.4618,
      "step": 747
    },
    {
      "epoch": 0.04921295458657499,
      "grad_norm": 0.8400937914848328,
      "learning_rate": 4.898285373291152e-05,
      "loss": 1.3205,
      "step": 748
    },
    {
      "epoch": 0.049278747306610526,
      "grad_norm": 0.9195184111595154,
      "learning_rate": 4.887580289089787e-05,
      "loss": 1.3075,
      "step": 749
    },
    {
      "epoch": 0.04934454002664605,
      "grad_norm": 1.210403561592102,
      "learning_rate": 4.876875720451511e-05,
      "loss": 1.3114,
      "step": 750
    },
    {
      "epoch": 0.04941033274668158,
      "grad_norm": 0.5409945845603943,
      "learning_rate": 4.866171716468074e-05,
      "loss": 1.5787,
      "step": 751
    },
    {
      "epoch": 0.049476125466717105,
      "grad_norm": 0.5661841034889221,
      "learning_rate": 4.855468326228638e-05,
      "loss": 1.5597,
      "step": 752
    },
    {
      "epoch": 0.04954191818675264,
      "grad_norm": 0.5858348608016968,
      "learning_rate": 4.8447655988195464e-05,
      "loss": 1.6057,
      "step": 753
    },
    {
      "epoch": 0.049607710906788165,
      "grad_norm": 0.5190932154655457,
      "learning_rate": 4.834063583324111e-05,
      "loss": 1.5158,
      "step": 754
    },
    {
      "epoch": 0.04967350362682369,
      "grad_norm": 0.6089287996292114,
      "learning_rate": 4.8233623288223704e-05,
      "loss": 1.7857,
      "step": 755
    },
    {
      "epoch": 0.04973929634685922,
      "grad_norm": 0.6427363753318787,
      "learning_rate": 4.8126618843908775e-05,
      "loss": 1.8208,
      "step": 756
    },
    {
      "epoch": 0.04980508906689475,
      "grad_norm": 0.5550252795219421,
      "learning_rate": 4.801962299102471e-05,
      "loss": 1.6078,
      "step": 757
    },
    {
      "epoch": 0.04987088178693028,
      "grad_norm": 0.6121551990509033,
      "learning_rate": 4.7912636220260473e-05,
      "loss": 1.725,
      "step": 758
    },
    {
      "epoch": 0.0499366745069658,
      "grad_norm": 0.6313748955726624,
      "learning_rate": 4.780565902226338e-05,
      "loss": 1.8692,
      "step": 759
    },
    {
      "epoch": 0.05000246722700133,
      "grad_norm": 0.5803863406181335,
      "learning_rate": 4.7698691887636854e-05,
      "loss": 1.5714,
      "step": 760
    },
    {
      "epoch": 0.05006825994703686,
      "grad_norm": 0.5760961174964905,
      "learning_rate": 4.759173530693814e-05,
      "loss": 1.6356,
      "step": 761
    },
    {
      "epoch": 0.05013405266707239,
      "grad_norm": 0.5896910429000854,
      "learning_rate": 4.748478977067614e-05,
      "loss": 1.5632,
      "step": 762
    },
    {
      "epoch": 0.050199845387107915,
      "grad_norm": 0.5783917903900146,
      "learning_rate": 4.737785576930908e-05,
      "loss": 1.531,
      "step": 763
    },
    {
      "epoch": 0.05026563810714344,
      "grad_norm": 0.6067524552345276,
      "learning_rate": 4.727093379324222e-05,
      "loss": 1.6955,
      "step": 764
    },
    {
      "epoch": 0.050331430827178975,
      "grad_norm": 0.5850797295570374,
      "learning_rate": 4.716402433282575e-05,
      "loss": 1.6401,
      "step": 765
    },
    {
      "epoch": 0.0503972235472145,
      "grad_norm": 0.6247797012329102,
      "learning_rate": 4.705712787835247e-05,
      "loss": 1.7098,
      "step": 766
    },
    {
      "epoch": 0.05046301626725003,
      "grad_norm": 0.5667417049407959,
      "learning_rate": 4.695024492005548e-05,
      "loss": 1.5379,
      "step": 767
    },
    {
      "epoch": 0.05052880898728556,
      "grad_norm": 0.5960890650749207,
      "learning_rate": 4.684337594810602e-05,
      "loss": 1.7614,
      "step": 768
    },
    {
      "epoch": 0.05059460170732109,
      "grad_norm": 0.6005357503890991,
      "learning_rate": 4.673652145261116e-05,
      "loss": 1.5532,
      "step": 769
    },
    {
      "epoch": 0.05066039442735661,
      "grad_norm": 0.6175205111503601,
      "learning_rate": 4.6629681923611603e-05,
      "loss": 1.6746,
      "step": 770
    },
    {
      "epoch": 0.05072618714739214,
      "grad_norm": 0.6387405395507812,
      "learning_rate": 4.652285785107943e-05,
      "loss": 1.6399,
      "step": 771
    },
    {
      "epoch": 0.05079197986742767,
      "grad_norm": 0.5912647247314453,
      "learning_rate": 4.64160497249158e-05,
      "loss": 1.7165,
      "step": 772
    },
    {
      "epoch": 0.0508577725874632,
      "grad_norm": 0.6159881353378296,
      "learning_rate": 4.630925803494877e-05,
      "loss": 1.6739,
      "step": 773
    },
    {
      "epoch": 0.050923565307498725,
      "grad_norm": 0.6138644218444824,
      "learning_rate": 4.6202483270931e-05,
      "loss": 1.535,
      "step": 774
    },
    {
      "epoch": 0.05098935802753425,
      "grad_norm": 0.6410723924636841,
      "learning_rate": 4.6095725922537533e-05,
      "loss": 1.6135,
      "step": 775
    },
    {
      "epoch": 0.051055150747569784,
      "grad_norm": 0.6804289817810059,
      "learning_rate": 4.598898647936354e-05,
      "loss": 1.639,
      "step": 776
    },
    {
      "epoch": 0.05112094346760531,
      "grad_norm": 0.6560742259025574,
      "learning_rate": 4.588226543092209e-05,
      "loss": 1.6589,
      "step": 777
    },
    {
      "epoch": 0.05118673618764084,
      "grad_norm": 0.6562034487724304,
      "learning_rate": 4.5775563266641894e-05,
      "loss": 1.5987,
      "step": 778
    },
    {
      "epoch": 0.05125252890767636,
      "grad_norm": 0.7900398969650269,
      "learning_rate": 4.566888047586507e-05,
      "loss": 1.8027,
      "step": 779
    },
    {
      "epoch": 0.051318321627711896,
      "grad_norm": 0.6825142502784729,
      "learning_rate": 4.556221754784482e-05,
      "loss": 1.5588,
      "step": 780
    },
    {
      "epoch": 0.05138411434774742,
      "grad_norm": 0.6759592294692993,
      "learning_rate": 4.545557497174331e-05,
      "loss": 1.535,
      "step": 781
    },
    {
      "epoch": 0.05144990706778295,
      "grad_norm": 0.7403655052185059,
      "learning_rate": 4.5348953236629395e-05,
      "loss": 1.6259,
      "step": 782
    },
    {
      "epoch": 0.051515699787818475,
      "grad_norm": 0.711732804775238,
      "learning_rate": 4.52423528314763e-05,
      "loss": 1.5692,
      "step": 783
    },
    {
      "epoch": 0.05158149250785401,
      "grad_norm": 0.6846985220909119,
      "learning_rate": 4.5135774245159454e-05,
      "loss": 1.5152,
      "step": 784
    },
    {
      "epoch": 0.051647285227889535,
      "grad_norm": 0.6845923066139221,
      "learning_rate": 4.502921796645424e-05,
      "loss": 1.5296,
      "step": 785
    },
    {
      "epoch": 0.05171307794792506,
      "grad_norm": 0.7074812054634094,
      "learning_rate": 4.492268448403369e-05,
      "loss": 1.4728,
      "step": 786
    },
    {
      "epoch": 0.05177887066796059,
      "grad_norm": 1.4424465894699097,
      "learning_rate": 4.4816174286466314e-05,
      "loss": 1.4922,
      "step": 787
    },
    {
      "epoch": 0.05184466338799612,
      "grad_norm": 0.7188639640808105,
      "learning_rate": 4.4709687862213866e-05,
      "loss": 1.6399,
      "step": 788
    },
    {
      "epoch": 0.05191045610803165,
      "grad_norm": 0.7116633057594299,
      "learning_rate": 4.4603225699629037e-05,
      "loss": 1.586,
      "step": 789
    },
    {
      "epoch": 0.05197624882806717,
      "grad_norm": 0.728759229183197,
      "learning_rate": 4.4496788286953266e-05,
      "loss": 1.4238,
      "step": 790
    },
    {
      "epoch": 0.0520420415481027,
      "grad_norm": 0.7237952947616577,
      "learning_rate": 4.439037611231448e-05,
      "loss": 1.514,
      "step": 791
    },
    {
      "epoch": 0.05210783426813823,
      "grad_norm": 0.7653865218162537,
      "learning_rate": 4.4283989663724875e-05,
      "loss": 1.6958,
      "step": 792
    },
    {
      "epoch": 0.05217362698817376,
      "grad_norm": 0.7304019331932068,
      "learning_rate": 4.4177629429078635e-05,
      "loss": 1.5366,
      "step": 793
    },
    {
      "epoch": 0.052239419708209285,
      "grad_norm": 0.7114179730415344,
      "learning_rate": 4.407129589614979e-05,
      "loss": 1.4206,
      "step": 794
    },
    {
      "epoch": 0.05230521242824481,
      "grad_norm": 0.7419894337654114,
      "learning_rate": 4.396498955258989e-05,
      "loss": 1.4686,
      "step": 795
    },
    {
      "epoch": 0.052371005148280345,
      "grad_norm": 0.8355379700660706,
      "learning_rate": 4.385871088592571e-05,
      "loss": 1.63,
      "step": 796
    },
    {
      "epoch": 0.05243679786831587,
      "grad_norm": 0.8412326574325562,
      "learning_rate": 4.3752460383557195e-05,
      "loss": 1.6383,
      "step": 797
    },
    {
      "epoch": 0.0525025905883514,
      "grad_norm": 0.8090490102767944,
      "learning_rate": 4.3646238532755114e-05,
      "loss": 1.407,
      "step": 798
    },
    {
      "epoch": 0.05256838330838693,
      "grad_norm": 0.8998046517372131,
      "learning_rate": 4.3540045820658804e-05,
      "loss": 1.625,
      "step": 799
    },
    {
      "epoch": 0.05263417602842246,
      "grad_norm": 1.3124433755874634,
      "learning_rate": 4.3433882734274e-05,
      "loss": 1.5433,
      "step": 800
    },
    {
      "epoch": 0.05269996874845798,
      "grad_norm": 0.5437705516815186,
      "learning_rate": 4.332774976047055e-05,
      "loss": 1.7119,
      "step": 801
    },
    {
      "epoch": 0.05276576146849351,
      "grad_norm": 0.5468183159828186,
      "learning_rate": 4.322164738598022e-05,
      "loss": 1.4418,
      "step": 802
    },
    {
      "epoch": 0.05283155418852904,
      "grad_norm": 0.5732606053352356,
      "learning_rate": 4.311557609739442e-05,
      "loss": 1.7286,
      "step": 803
    },
    {
      "epoch": 0.05289734690856457,
      "grad_norm": 0.5762453675270081,
      "learning_rate": 4.300953638116204e-05,
      "loss": 1.7308,
      "step": 804
    },
    {
      "epoch": 0.052963139628600095,
      "grad_norm": 0.5938736796379089,
      "learning_rate": 4.290352872358714e-05,
      "loss": 1.7055,
      "step": 805
    },
    {
      "epoch": 0.05302893234863562,
      "grad_norm": 0.5911692976951599,
      "learning_rate": 4.27975536108268e-05,
      "loss": 1.631,
      "step": 806
    },
    {
      "epoch": 0.053094725068671154,
      "grad_norm": 0.5975512266159058,
      "learning_rate": 4.2691611528888775e-05,
      "loss": 1.6773,
      "step": 807
    },
    {
      "epoch": 0.05316051778870668,
      "grad_norm": 0.5482115745544434,
      "learning_rate": 4.258570296362942e-05,
      "loss": 1.5443,
      "step": 808
    },
    {
      "epoch": 0.05322631050874221,
      "grad_norm": 0.6074912548065186,
      "learning_rate": 4.24798284007513e-05,
      "loss": 1.6919,
      "step": 809
    },
    {
      "epoch": 0.05329210322877773,
      "grad_norm": 0.5921036601066589,
      "learning_rate": 4.2373988325801145e-05,
      "loss": 1.6611,
      "step": 810
    },
    {
      "epoch": 0.053357895948813266,
      "grad_norm": 0.641273558139801,
      "learning_rate": 4.2268183224167456e-05,
      "loss": 1.8859,
      "step": 811
    },
    {
      "epoch": 0.05342368866884879,
      "grad_norm": 0.610242486000061,
      "learning_rate": 4.216241358107831e-05,
      "loss": 1.5211,
      "step": 812
    },
    {
      "epoch": 0.05348948138888432,
      "grad_norm": 0.6088224053382874,
      "learning_rate": 4.205667988159921e-05,
      "loss": 1.6132,
      "step": 813
    },
    {
      "epoch": 0.053555274108919845,
      "grad_norm": 0.5857619643211365,
      "learning_rate": 4.195098261063087e-05,
      "loss": 1.5505,
      "step": 814
    },
    {
      "epoch": 0.05362106682895538,
      "grad_norm": 0.6356807351112366,
      "learning_rate": 4.1845322252906864e-05,
      "loss": 1.6472,
      "step": 815
    },
    {
      "epoch": 0.053686859548990905,
      "grad_norm": 0.5856146216392517,
      "learning_rate": 4.173969929299151e-05,
      "loss": 1.7503,
      "step": 816
    },
    {
      "epoch": 0.05375265226902643,
      "grad_norm": 0.60245680809021,
      "learning_rate": 4.1634114215277625e-05,
      "loss": 1.5542,
      "step": 817
    },
    {
      "epoch": 0.05381844498906196,
      "grad_norm": 0.6490529179573059,
      "learning_rate": 4.152856750398426e-05,
      "loss": 1.5974,
      "step": 818
    },
    {
      "epoch": 0.05388423770909749,
      "grad_norm": 0.6497132778167725,
      "learning_rate": 4.1423059643154564e-05,
      "loss": 1.603,
      "step": 819
    },
    {
      "epoch": 0.05395003042913302,
      "grad_norm": 0.5978454351425171,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.4747,
      "step": 820
    },
    {
      "epoch": 0.05401582314916854,
      "grad_norm": 0.6695238947868347,
      "learning_rate": 4.1212162408165595e-05,
      "loss": 1.6682,
      "step": 821
    },
    {
      "epoch": 0.05408161586920407,
      "grad_norm": 0.6585442423820496,
      "learning_rate": 4.110677400119285e-05,
      "loss": 1.72,
      "step": 822
    },
    {
      "epoch": 0.0541474085892396,
      "grad_norm": 0.6088001132011414,
      "learning_rate": 4.100142637905238e-05,
      "loss": 1.474,
      "step": 823
    },
    {
      "epoch": 0.05421320130927513,
      "grad_norm": 0.6855148673057556,
      "learning_rate": 4.0896120024874286e-05,
      "loss": 1.6749,
      "step": 824
    },
    {
      "epoch": 0.054278994029310655,
      "grad_norm": 0.6727811694145203,
      "learning_rate": 4.079085542159938e-05,
      "loss": 1.7432,
      "step": 825
    },
    {
      "epoch": 0.05434478674934618,
      "grad_norm": 0.6346832513809204,
      "learning_rate": 4.068563305197706e-05,
      "loss": 1.5826,
      "step": 826
    },
    {
      "epoch": 0.054410579469381715,
      "grad_norm": 0.6359438300132751,
      "learning_rate": 4.0580453398563e-05,
      "loss": 1.5543,
      "step": 827
    },
    {
      "epoch": 0.05447637218941724,
      "grad_norm": 0.6643136739730835,
      "learning_rate": 4.047531694371695e-05,
      "loss": 1.5587,
      "step": 828
    },
    {
      "epoch": 0.05454216490945277,
      "grad_norm": 0.6467865109443665,
      "learning_rate": 4.037022416960058e-05,
      "loss": 1.7538,
      "step": 829
    },
    {
      "epoch": 0.0546079576294883,
      "grad_norm": 0.6873615980148315,
      "learning_rate": 4.0265175558175265e-05,
      "loss": 1.5784,
      "step": 830
    },
    {
      "epoch": 0.05467375034952383,
      "grad_norm": 0.6892521977424622,
      "learning_rate": 4.01601715911998e-05,
      "loss": 1.5759,
      "step": 831
    },
    {
      "epoch": 0.05473954306955935,
      "grad_norm": 0.6923861503601074,
      "learning_rate": 4.005521275022826e-05,
      "loss": 1.6928,
      "step": 832
    },
    {
      "epoch": 0.05480533578959488,
      "grad_norm": 0.9242671132087708,
      "learning_rate": 3.9950299516607766e-05,
      "loss": 1.7272,
      "step": 833
    },
    {
      "epoch": 0.05487112850963041,
      "grad_norm": 0.7108779549598694,
      "learning_rate": 3.9845432371476264e-05,
      "loss": 1.632,
      "step": 834
    },
    {
      "epoch": 0.05493692122966594,
      "grad_norm": 0.7247679233551025,
      "learning_rate": 3.9740611795760376e-05,
      "loss": 1.5377,
      "step": 835
    },
    {
      "epoch": 0.055002713949701465,
      "grad_norm": 0.7193149328231812,
      "learning_rate": 3.9635838270173107e-05,
      "loss": 1.431,
      "step": 836
    },
    {
      "epoch": 0.05506850666973699,
      "grad_norm": 0.7430593967437744,
      "learning_rate": 3.9531112275211736e-05,
      "loss": 1.5027,
      "step": 837
    },
    {
      "epoch": 0.055134299389772524,
      "grad_norm": 0.7426053285598755,
      "learning_rate": 3.9426434291155526e-05,
      "loss": 1.6552,
      "step": 838
    },
    {
      "epoch": 0.05520009210980805,
      "grad_norm": 0.7447395324707031,
      "learning_rate": 3.9321804798063565e-05,
      "loss": 1.5632,
      "step": 839
    },
    {
      "epoch": 0.05526588482984358,
      "grad_norm": 0.7917671203613281,
      "learning_rate": 3.921722427577257e-05,
      "loss": 1.5141,
      "step": 840
    },
    {
      "epoch": 0.0553316775498791,
      "grad_norm": 0.815054178237915,
      "learning_rate": 3.9112693203894664e-05,
      "loss": 1.5968,
      "step": 841
    },
    {
      "epoch": 0.055397470269914637,
      "grad_norm": 0.8161969184875488,
      "learning_rate": 3.900821206181521e-05,
      "loss": 1.487,
      "step": 842
    },
    {
      "epoch": 0.05546326298995016,
      "grad_norm": 0.7944288849830627,
      "learning_rate": 3.890378132869059e-05,
      "loss": 1.6617,
      "step": 843
    },
    {
      "epoch": 0.05552905570998569,
      "grad_norm": 0.801561176776886,
      "learning_rate": 3.879940148344595e-05,
      "loss": 1.4835,
      "step": 844
    },
    {
      "epoch": 0.055594848430021215,
      "grad_norm": 0.8297501802444458,
      "learning_rate": 3.8695073004773106e-05,
      "loss": 1.3569,
      "step": 845
    },
    {
      "epoch": 0.05566064115005675,
      "grad_norm": 0.8283421993255615,
      "learning_rate": 3.859079637112833e-05,
      "loss": 1.4448,
      "step": 846
    },
    {
      "epoch": 0.055726433870092275,
      "grad_norm": 0.862443745136261,
      "learning_rate": 3.84865720607301e-05,
      "loss": 1.558,
      "step": 847
    },
    {
      "epoch": 0.0557922265901278,
      "grad_norm": 0.8980772495269775,
      "learning_rate": 3.838240055155692e-05,
      "loss": 1.4493,
      "step": 848
    },
    {
      "epoch": 0.05585801931016333,
      "grad_norm": 0.9152320623397827,
      "learning_rate": 3.827828232134519e-05,
      "loss": 1.2151,
      "step": 849
    },
    {
      "epoch": 0.05592381203019886,
      "grad_norm": 1.169055700302124,
      "learning_rate": 3.8174217847586904e-05,
      "loss": 1.2846,
      "step": 850
    },
    {
      "epoch": 0.05598960475023439,
      "grad_norm": 0.5100705623626709,
      "learning_rate": 3.8070207607527584e-05,
      "loss": 1.5764,
      "step": 851
    },
    {
      "epoch": 0.05605539747026991,
      "grad_norm": 0.5505961179733276,
      "learning_rate": 3.796625207816401e-05,
      "loss": 1.7925,
      "step": 852
    },
    {
      "epoch": 0.05612119019030544,
      "grad_norm": 0.6031649708747864,
      "learning_rate": 3.7862351736242066e-05,
      "loss": 1.7848,
      "step": 853
    },
    {
      "epoch": 0.05618698291034097,
      "grad_norm": 0.6231903433799744,
      "learning_rate": 3.775850705825454e-05,
      "loss": 1.7385,
      "step": 854
    },
    {
      "epoch": 0.0562527756303765,
      "grad_norm": 0.5796992778778076,
      "learning_rate": 3.765471852043892e-05,
      "loss": 1.6627,
      "step": 855
    },
    {
      "epoch": 0.056318568350412025,
      "grad_norm": 0.5962229371070862,
      "learning_rate": 3.755098659877527e-05,
      "loss": 1.7171,
      "step": 856
    },
    {
      "epoch": 0.05638436107044756,
      "grad_norm": 0.6059638857841492,
      "learning_rate": 3.7447311768983964e-05,
      "loss": 1.6715,
      "step": 857
    },
    {
      "epoch": 0.056450153790483085,
      "grad_norm": 0.6489062905311584,
      "learning_rate": 3.7343694506523605e-05,
      "loss": 1.7469,
      "step": 858
    },
    {
      "epoch": 0.05651594651051861,
      "grad_norm": 0.6001795530319214,
      "learning_rate": 3.7240135286588765e-05,
      "loss": 1.6136,
      "step": 859
    },
    {
      "epoch": 0.05658173923055414,
      "grad_norm": 0.6038196086883545,
      "learning_rate": 3.713663458410779e-05,
      "loss": 1.6115,
      "step": 860
    },
    {
      "epoch": 0.05664753195058967,
      "grad_norm": 0.584757924079895,
      "learning_rate": 3.703319287374069e-05,
      "loss": 1.5293,
      "step": 861
    },
    {
      "epoch": 0.0567133246706252,
      "grad_norm": 0.6052143573760986,
      "learning_rate": 3.692981062987698e-05,
      "loss": 1.8357,
      "step": 862
    },
    {
      "epoch": 0.05677911739066072,
      "grad_norm": 0.6260353326797485,
      "learning_rate": 3.682648832663339e-05,
      "loss": 1.6054,
      "step": 863
    },
    {
      "epoch": 0.05684491011069625,
      "grad_norm": 0.643218457698822,
      "learning_rate": 3.67232264378518e-05,
      "loss": 1.7711,
      "step": 864
    },
    {
      "epoch": 0.05691070283073178,
      "grad_norm": 0.577084481716156,
      "learning_rate": 3.662002543709702e-05,
      "loss": 1.582,
      "step": 865
    },
    {
      "epoch": 0.05697649555076731,
      "grad_norm": 0.7629356384277344,
      "learning_rate": 3.6516885797654594e-05,
      "loss": 1.6112,
      "step": 866
    },
    {
      "epoch": 0.057042288270802835,
      "grad_norm": 0.6926255226135254,
      "learning_rate": 3.641380799252869e-05,
      "loss": 1.537,
      "step": 867
    },
    {
      "epoch": 0.05710808099083836,
      "grad_norm": 0.6118739247322083,
      "learning_rate": 3.6310792494439894e-05,
      "loss": 1.6167,
      "step": 868
    },
    {
      "epoch": 0.057173873710873895,
      "grad_norm": 0.6357191205024719,
      "learning_rate": 3.620783977582305e-05,
      "loss": 1.4908,
      "step": 869
    },
    {
      "epoch": 0.05723966643090942,
      "grad_norm": 0.6403725743293762,
      "learning_rate": 3.61049503088251e-05,
      "loss": 1.469,
      "step": 870
    },
    {
      "epoch": 0.05730545915094495,
      "grad_norm": 0.6135770678520203,
      "learning_rate": 3.6002124565302875e-05,
      "loss": 1.4758,
      "step": 871
    },
    {
      "epoch": 0.05737125187098047,
      "grad_norm": 0.6306059956550598,
      "learning_rate": 3.5899363016821e-05,
      "loss": 1.5523,
      "step": 872
    },
    {
      "epoch": 0.05743704459101601,
      "grad_norm": 0.6534183025360107,
      "learning_rate": 3.579666613464968e-05,
      "loss": 1.5653,
      "step": 873
    },
    {
      "epoch": 0.05750283731105153,
      "grad_norm": 0.6130310893058777,
      "learning_rate": 3.56940343897626e-05,
      "loss": 1.547,
      "step": 874
    },
    {
      "epoch": 0.05756863003108706,
      "grad_norm": 0.6713092923164368,
      "learning_rate": 3.559146825283465e-05,
      "loss": 1.6401,
      "step": 875
    },
    {
      "epoch": 0.057634422751122585,
      "grad_norm": 0.6866581439971924,
      "learning_rate": 3.5488968194239926e-05,
      "loss": 1.6317,
      "step": 876
    },
    {
      "epoch": 0.05770021547115812,
      "grad_norm": 0.6544456481933594,
      "learning_rate": 3.538653468404936e-05,
      "loss": 1.5999,
      "step": 877
    },
    {
      "epoch": 0.057766008191193645,
      "grad_norm": 0.6867201328277588,
      "learning_rate": 3.528416819202881e-05,
      "loss": 1.6641,
      "step": 878
    },
    {
      "epoch": 0.05783180091122917,
      "grad_norm": 0.7083088159561157,
      "learning_rate": 3.5181869187636735e-05,
      "loss": 1.499,
      "step": 879
    },
    {
      "epoch": 0.0578975936312647,
      "grad_norm": 0.7546277046203613,
      "learning_rate": 3.5079638140022094e-05,
      "loss": 1.6342,
      "step": 880
    },
    {
      "epoch": 0.05796338635130023,
      "grad_norm": 0.7427401542663574,
      "learning_rate": 3.497747551802221e-05,
      "loss": 1.7578,
      "step": 881
    },
    {
      "epoch": 0.05802917907133576,
      "grad_norm": 0.7476483583450317,
      "learning_rate": 3.487538179016054e-05,
      "loss": 1.5637,
      "step": 882
    },
    {
      "epoch": 0.05809497179137128,
      "grad_norm": 0.7196775078773499,
      "learning_rate": 3.4773357424644685e-05,
      "loss": 1.5247,
      "step": 883
    },
    {
      "epoch": 0.05816076451140681,
      "grad_norm": 0.7455043792724609,
      "learning_rate": 3.467140288936407e-05,
      "loss": 1.5444,
      "step": 884
    },
    {
      "epoch": 0.05822655723144234,
      "grad_norm": 0.7095162868499756,
      "learning_rate": 3.456951865188791e-05,
      "loss": 1.6029,
      "step": 885
    },
    {
      "epoch": 0.05829234995147787,
      "grad_norm": 0.7810092568397522,
      "learning_rate": 3.446770517946303e-05,
      "loss": 1.5827,
      "step": 886
    },
    {
      "epoch": 0.058358142671513395,
      "grad_norm": 0.7805618643760681,
      "learning_rate": 3.43659629390117e-05,
      "loss": 1.562,
      "step": 887
    },
    {
      "epoch": 0.05842393539154893,
      "grad_norm": 0.7464081048965454,
      "learning_rate": 3.426429239712952e-05,
      "loss": 1.646,
      "step": 888
    },
    {
      "epoch": 0.058489728111584455,
      "grad_norm": 0.7130300998687744,
      "learning_rate": 3.4162694020083294e-05,
      "loss": 1.3213,
      "step": 889
    },
    {
      "epoch": 0.05855552083161998,
      "grad_norm": 0.7412570118904114,
      "learning_rate": 3.406116827380889e-05,
      "loss": 1.6022,
      "step": 890
    },
    {
      "epoch": 0.05862131355165551,
      "grad_norm": 0.7365239262580872,
      "learning_rate": 3.395971562390905e-05,
      "loss": 1.5959,
      "step": 891
    },
    {
      "epoch": 0.05868710627169104,
      "grad_norm": 0.8533955812454224,
      "learning_rate": 3.385833653565132e-05,
      "loss": 1.766,
      "step": 892
    },
    {
      "epoch": 0.05875289899172657,
      "grad_norm": 0.8184581995010376,
      "learning_rate": 3.375703147396583e-05,
      "loss": 1.5165,
      "step": 893
    },
    {
      "epoch": 0.05881869171176209,
      "grad_norm": 0.7530040740966797,
      "learning_rate": 3.365580090344329e-05,
      "loss": 1.4796,
      "step": 894
    },
    {
      "epoch": 0.05888448443179762,
      "grad_norm": 0.8272554278373718,
      "learning_rate": 3.3554645288332775e-05,
      "loss": 1.3817,
      "step": 895
    },
    {
      "epoch": 0.05895027715183315,
      "grad_norm": 0.8762139678001404,
      "learning_rate": 3.345356509253959e-05,
      "loss": 1.4379,
      "step": 896
    },
    {
      "epoch": 0.05901606987186868,
      "grad_norm": 0.9064789414405823,
      "learning_rate": 3.335256077962317e-05,
      "loss": 1.5902,
      "step": 897
    },
    {
      "epoch": 0.059081862591904205,
      "grad_norm": 0.8573328852653503,
      "learning_rate": 3.325163281279494e-05,
      "loss": 1.4975,
      "step": 898
    },
    {
      "epoch": 0.05914765531193973,
      "grad_norm": 0.9656907320022583,
      "learning_rate": 3.315078165491622e-05,
      "loss": 1.5053,
      "step": 899
    },
    {
      "epoch": 0.059213448031975265,
      "grad_norm": 1.2840112447738647,
      "learning_rate": 3.305000776849606e-05,
      "loss": 1.2657,
      "step": 900
    },
    {
      "epoch": 0.05927924075201079,
      "grad_norm": 0.5296594500541687,
      "learning_rate": 3.294931161568914e-05,
      "loss": 1.7877,
      "step": 901
    },
    {
      "epoch": 0.05934503347204632,
      "grad_norm": 0.5523350834846497,
      "learning_rate": 3.2848693658293675e-05,
      "loss": 1.6165,
      "step": 902
    },
    {
      "epoch": 0.05941082619208184,
      "grad_norm": 0.5559424161911011,
      "learning_rate": 3.274815435774921e-05,
      "loss": 1.6415,
      "step": 903
    },
    {
      "epoch": 0.05947661891211738,
      "grad_norm": 0.6036515831947327,
      "learning_rate": 3.264769417513463e-05,
      "loss": 1.6959,
      "step": 904
    },
    {
      "epoch": 0.0595424116321529,
      "grad_norm": 0.5790659189224243,
      "learning_rate": 3.254731357116597e-05,
      "loss": 1.6368,
      "step": 905
    },
    {
      "epoch": 0.05960820435218843,
      "grad_norm": 0.5860310792922974,
      "learning_rate": 3.244701300619427e-05,
      "loss": 1.5645,
      "step": 906
    },
    {
      "epoch": 0.059673997072223955,
      "grad_norm": 0.5953836441040039,
      "learning_rate": 3.2346792940203554e-05,
      "loss": 1.5427,
      "step": 907
    },
    {
      "epoch": 0.05973978979225949,
      "grad_norm": 0.574989378452301,
      "learning_rate": 3.224665383280867e-05,
      "loss": 1.7678,
      "step": 908
    },
    {
      "epoch": 0.059805582512295015,
      "grad_norm": 0.5947942733764648,
      "learning_rate": 3.214659614325313e-05,
      "loss": 1.5293,
      "step": 909
    },
    {
      "epoch": 0.05987137523233054,
      "grad_norm": 0.5816536545753479,
      "learning_rate": 3.2046620330407116e-05,
      "loss": 1.4343,
      "step": 910
    },
    {
      "epoch": 0.05993716795236607,
      "grad_norm": 0.5763691663742065,
      "learning_rate": 3.194672685276532e-05,
      "loss": 1.6881,
      "step": 911
    },
    {
      "epoch": 0.0600029606724016,
      "grad_norm": 0.629294753074646,
      "learning_rate": 3.1846916168444806e-05,
      "loss": 1.7855,
      "step": 912
    },
    {
      "epoch": 0.06006875339243713,
      "grad_norm": 0.6405010223388672,
      "learning_rate": 3.174718873518298e-05,
      "loss": 1.6608,
      "step": 913
    },
    {
      "epoch": 0.06013454611247265,
      "grad_norm": 0.5852246284484863,
      "learning_rate": 3.16475450103354e-05,
      "loss": 1.6286,
      "step": 914
    },
    {
      "epoch": 0.06020033883250818,
      "grad_norm": 0.6519684195518494,
      "learning_rate": 3.154798545087379e-05,
      "loss": 1.6609,
      "step": 915
    },
    {
      "epoch": 0.06026613155254371,
      "grad_norm": 0.6138909459114075,
      "learning_rate": 3.144851051338385e-05,
      "loss": 1.7577,
      "step": 916
    },
    {
      "epoch": 0.06033192427257924,
      "grad_norm": 0.5902788639068604,
      "learning_rate": 3.1349120654063225e-05,
      "loss": 1.622,
      "step": 917
    },
    {
      "epoch": 0.060397716992614765,
      "grad_norm": 0.6420292854309082,
      "learning_rate": 3.124981632871937e-05,
      "loss": 1.6161,
      "step": 918
    },
    {
      "epoch": 0.0604635097126503,
      "grad_norm": 0.6139567494392395,
      "learning_rate": 3.115059799276748e-05,
      "loss": 1.682,
      "step": 919
    },
    {
      "epoch": 0.060529302432685825,
      "grad_norm": 0.6504656672477722,
      "learning_rate": 3.1051466101228385e-05,
      "loss": 1.5961,
      "step": 920
    },
    {
      "epoch": 0.06059509515272135,
      "grad_norm": 0.6670247912406921,
      "learning_rate": 3.09524211087265e-05,
      "loss": 1.6288,
      "step": 921
    },
    {
      "epoch": 0.06066088787275688,
      "grad_norm": 0.6252246499061584,
      "learning_rate": 3.08534634694877e-05,
      "loss": 1.527,
      "step": 922
    },
    {
      "epoch": 0.06072668059279241,
      "grad_norm": 0.6155936121940613,
      "learning_rate": 3.0754593637337276e-05,
      "loss": 1.5781,
      "step": 923
    },
    {
      "epoch": 0.06079247331282794,
      "grad_norm": 0.6207473278045654,
      "learning_rate": 3.065581206569782e-05,
      "loss": 1.3617,
      "step": 924
    },
    {
      "epoch": 0.06085826603286346,
      "grad_norm": 0.6116368174552917,
      "learning_rate": 3.055711920758709e-05,
      "loss": 1.6183,
      "step": 925
    },
    {
      "epoch": 0.06092405875289899,
      "grad_norm": 0.646839439868927,
      "learning_rate": 3.0458515515616115e-05,
      "loss": 1.5567,
      "step": 926
    },
    {
      "epoch": 0.06098985147293452,
      "grad_norm": 0.6746354103088379,
      "learning_rate": 3.036000144198693e-05,
      "loss": 1.5467,
      "step": 927
    },
    {
      "epoch": 0.06105564419297005,
      "grad_norm": 0.639434278011322,
      "learning_rate": 3.0261577438490585e-05,
      "loss": 1.5425,
      "step": 928
    },
    {
      "epoch": 0.061121436913005575,
      "grad_norm": 0.685247540473938,
      "learning_rate": 3.0163243956505095e-05,
      "loss": 1.6157,
      "step": 929
    },
    {
      "epoch": 0.0611872296330411,
      "grad_norm": 0.6764075756072998,
      "learning_rate": 3.006500144699328e-05,
      "loss": 1.5882,
      "step": 930
    },
    {
      "epoch": 0.061253022353076635,
      "grad_norm": 0.6915985345840454,
      "learning_rate": 2.99668503605008e-05,
      "loss": 1.51,
      "step": 931
    },
    {
      "epoch": 0.06131881507311216,
      "grad_norm": 0.74942946434021,
      "learning_rate": 2.986879114715403e-05,
      "loss": 1.4684,
      "step": 932
    },
    {
      "epoch": 0.06138460779314769,
      "grad_norm": 0.6873716711997986,
      "learning_rate": 2.977082425665802e-05,
      "loss": 1.6268,
      "step": 933
    },
    {
      "epoch": 0.06145040051318321,
      "grad_norm": 0.7743644714355469,
      "learning_rate": 2.967295013829442e-05,
      "loss": 1.564,
      "step": 934
    },
    {
      "epoch": 0.06151619323321875,
      "grad_norm": 0.702805757522583,
      "learning_rate": 2.95751692409194e-05,
      "loss": 1.7741,
      "step": 935
    },
    {
      "epoch": 0.06158198595325427,
      "grad_norm": 0.7666968107223511,
      "learning_rate": 2.947748201296163e-05,
      "loss": 1.6298,
      "step": 936
    },
    {
      "epoch": 0.0616477786732898,
      "grad_norm": 0.700221061706543,
      "learning_rate": 2.9379888902420215e-05,
      "loss": 1.5275,
      "step": 937
    },
    {
      "epoch": 0.061713571393325325,
      "grad_norm": 0.7477214932441711,
      "learning_rate": 2.9282390356862606e-05,
      "loss": 1.5412,
      "step": 938
    },
    {
      "epoch": 0.06177936411336086,
      "grad_norm": 0.7472383975982666,
      "learning_rate": 2.9184986823422623e-05,
      "loss": 1.547,
      "step": 939
    },
    {
      "epoch": 0.061845156833396385,
      "grad_norm": 0.7094493508338928,
      "learning_rate": 2.9087678748798297e-05,
      "loss": 1.4617,
      "step": 940
    },
    {
      "epoch": 0.06191094955343191,
      "grad_norm": 0.7252383828163147,
      "learning_rate": 2.899046657924992e-05,
      "loss": 1.3699,
      "step": 941
    },
    {
      "epoch": 0.06197674227346744,
      "grad_norm": 0.7631134390830994,
      "learning_rate": 2.8893350760597924e-05,
      "loss": 1.567,
      "step": 942
    },
    {
      "epoch": 0.06204253499350297,
      "grad_norm": 0.7023096680641174,
      "learning_rate": 2.8796331738220912e-05,
      "loss": 1.4324,
      "step": 943
    },
    {
      "epoch": 0.0621083277135385,
      "grad_norm": 0.7988854050636292,
      "learning_rate": 2.8699409957053535e-05,
      "loss": 1.6788,
      "step": 944
    },
    {
      "epoch": 0.06217412043357402,
      "grad_norm": 0.8351010084152222,
      "learning_rate": 2.860258586158454e-05,
      "loss": 1.6451,
      "step": 945
    },
    {
      "epoch": 0.062239913153609557,
      "grad_norm": 0.8499585390090942,
      "learning_rate": 2.8505859895854604e-05,
      "loss": 1.6087,
      "step": 946
    },
    {
      "epoch": 0.06230570587364508,
      "grad_norm": 0.9862839579582214,
      "learning_rate": 2.840923250345442e-05,
      "loss": 1.5487,
      "step": 947
    },
    {
      "epoch": 0.06237149859368061,
      "grad_norm": 0.9058715105056763,
      "learning_rate": 2.8312704127522655e-05,
      "loss": 1.3746,
      "step": 948
    },
    {
      "epoch": 0.062437291313716135,
      "grad_norm": 0.9685963988304138,
      "learning_rate": 2.821627521074383e-05,
      "loss": 1.4866,
      "step": 949
    },
    {
      "epoch": 0.06250308403375167,
      "grad_norm": 1.311897873878479,
      "learning_rate": 2.8119946195346375e-05,
      "loss": 1.5507,
      "step": 950
    },
    {
      "epoch": 0.0625688767537872,
      "grad_norm": 0.5097134113311768,
      "learning_rate": 2.8023717523100508e-05,
      "loss": 1.8342,
      "step": 951
    },
    {
      "epoch": 0.06263466947382272,
      "grad_norm": 0.5375131368637085,
      "learning_rate": 2.792758963531632e-05,
      "loss": 1.6841,
      "step": 952
    },
    {
      "epoch": 0.06270046219385825,
      "grad_norm": 0.6458749771118164,
      "learning_rate": 2.7831562972841696e-05,
      "loss": 2.0434,
      "step": 953
    },
    {
      "epoch": 0.06276625491389377,
      "grad_norm": 0.5497531890869141,
      "learning_rate": 2.773563797606028e-05,
      "loss": 1.6057,
      "step": 954
    },
    {
      "epoch": 0.0628320476339293,
      "grad_norm": 0.5604535341262817,
      "learning_rate": 2.7639815084889476e-05,
      "loss": 1.6338,
      "step": 955
    },
    {
      "epoch": 0.06289784035396484,
      "grad_norm": 0.6276973485946655,
      "learning_rate": 2.7544094738778436e-05,
      "loss": 1.7273,
      "step": 956
    },
    {
      "epoch": 0.06296363307400037,
      "grad_norm": 0.6032599210739136,
      "learning_rate": 2.744847737670601e-05,
      "loss": 1.6246,
      "step": 957
    },
    {
      "epoch": 0.06302942579403589,
      "grad_norm": 0.6191315650939941,
      "learning_rate": 2.7352963437178786e-05,
      "loss": 1.6568,
      "step": 958
    },
    {
      "epoch": 0.06309521851407142,
      "grad_norm": 0.5911609530448914,
      "learning_rate": 2.7257553358229034e-05,
      "loss": 1.6472,
      "step": 959
    },
    {
      "epoch": 0.06316101123410695,
      "grad_norm": 0.5993046164512634,
      "learning_rate": 2.7162247577412715e-05,
      "loss": 1.5908,
      "step": 960
    },
    {
      "epoch": 0.06322680395414247,
      "grad_norm": 0.6404312252998352,
      "learning_rate": 2.7067046531807494e-05,
      "loss": 1.6952,
      "step": 961
    },
    {
      "epoch": 0.063292596674178,
      "grad_norm": 0.6177670359611511,
      "learning_rate": 2.6971950658010666e-05,
      "loss": 1.5803,
      "step": 962
    },
    {
      "epoch": 0.06335838939421352,
      "grad_norm": 0.6164083480834961,
      "learning_rate": 2.6876960392137217e-05,
      "loss": 1.5917,
      "step": 963
    },
    {
      "epoch": 0.06342418211424906,
      "grad_norm": 0.6648538708686829,
      "learning_rate": 2.678207616981787e-05,
      "loss": 1.6052,
      "step": 964
    },
    {
      "epoch": 0.06348997483428459,
      "grad_norm": 0.6657065153121948,
      "learning_rate": 2.6687298426196973e-05,
      "loss": 1.7399,
      "step": 965
    },
    {
      "epoch": 0.06355576755432012,
      "grad_norm": 0.6626632213592529,
      "learning_rate": 2.6592627595930542e-05,
      "loss": 1.6635,
      "step": 966
    },
    {
      "epoch": 0.06362156027435564,
      "grad_norm": 0.6235553622245789,
      "learning_rate": 2.6498064113184338e-05,
      "loss": 1.6467,
      "step": 967
    },
    {
      "epoch": 0.06368735299439117,
      "grad_norm": 0.6630668044090271,
      "learning_rate": 2.6403608411631742e-05,
      "loss": 1.6262,
      "step": 968
    },
    {
      "epoch": 0.0637531457144267,
      "grad_norm": 0.6901178956031799,
      "learning_rate": 2.6309260924451907e-05,
      "loss": 1.687,
      "step": 969
    },
    {
      "epoch": 0.06381893843446222,
      "grad_norm": 0.6554874777793884,
      "learning_rate": 2.62150220843277e-05,
      "loss": 1.7402,
      "step": 970
    },
    {
      "epoch": 0.06388473115449775,
      "grad_norm": 0.6539959907531738,
      "learning_rate": 2.612089232344371e-05,
      "loss": 1.7455,
      "step": 971
    },
    {
      "epoch": 0.06395052387453329,
      "grad_norm": 0.6740013360977173,
      "learning_rate": 2.602687207348429e-05,
      "loss": 1.6105,
      "step": 972
    },
    {
      "epoch": 0.06401631659456881,
      "grad_norm": 0.6463841199874878,
      "learning_rate": 2.593296176563157e-05,
      "loss": 1.5821,
      "step": 973
    },
    {
      "epoch": 0.06408210931460434,
      "grad_norm": 0.6570295691490173,
      "learning_rate": 2.5839161830563474e-05,
      "loss": 1.6883,
      "step": 974
    },
    {
      "epoch": 0.06414790203463987,
      "grad_norm": 0.6713987588882446,
      "learning_rate": 2.5745472698451767e-05,
      "loss": 1.5053,
      "step": 975
    },
    {
      "epoch": 0.0642136947546754,
      "grad_norm": 0.6906765103340149,
      "learning_rate": 2.565189479896005e-05,
      "loss": 1.5105,
      "step": 976
    },
    {
      "epoch": 0.06427948747471092,
      "grad_norm": 0.6981975436210632,
      "learning_rate": 2.555842856124182e-05,
      "loss": 1.5976,
      "step": 977
    },
    {
      "epoch": 0.06434528019474645,
      "grad_norm": 0.7114136815071106,
      "learning_rate": 2.546507441393845e-05,
      "loss": 1.6393,
      "step": 978
    },
    {
      "epoch": 0.06441107291478199,
      "grad_norm": 0.6741265058517456,
      "learning_rate": 2.5371832785177273e-05,
      "loss": 1.6601,
      "step": 979
    },
    {
      "epoch": 0.06447686563481751,
      "grad_norm": 0.7347429990768433,
      "learning_rate": 2.5278704102569662e-05,
      "loss": 1.6784,
      "step": 980
    },
    {
      "epoch": 0.06454265835485304,
      "grad_norm": 0.6604593992233276,
      "learning_rate": 2.518568879320895e-05,
      "loss": 1.5858,
      "step": 981
    },
    {
      "epoch": 0.06460845107488856,
      "grad_norm": 0.6967578530311584,
      "learning_rate": 2.509278728366855e-05,
      "loss": 1.5591,
      "step": 982
    },
    {
      "epoch": 0.06467424379492409,
      "grad_norm": 0.744243323802948,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.5185,
      "step": 983
    },
    {
      "epoch": 0.06474003651495962,
      "grad_norm": 0.7131791114807129,
      "learning_rate": 2.4907327367730963e-05,
      "loss": 1.5966,
      "step": 984
    },
    {
      "epoch": 0.06480582923499514,
      "grad_norm": 0.745911180973053,
      "learning_rate": 2.4814769811863313e-05,
      "loss": 1.6642,
      "step": 985
    },
    {
      "epoch": 0.06487162195503067,
      "grad_norm": 0.7129960656166077,
      "learning_rate": 2.472232775687119e-05,
      "loss": 1.4803,
      "step": 986
    },
    {
      "epoch": 0.06493741467506621,
      "grad_norm": 0.7604677677154541,
      "learning_rate": 2.463000162669903e-05,
      "loss": 1.4905,
      "step": 987
    },
    {
      "epoch": 0.06500320739510174,
      "grad_norm": 0.7828547358512878,
      "learning_rate": 2.453779184475964e-05,
      "loss": 1.5679,
      "step": 988
    },
    {
      "epoch": 0.06506900011513726,
      "grad_norm": 0.7184737324714661,
      "learning_rate": 2.4445698833932234e-05,
      "loss": 1.4162,
      "step": 989
    },
    {
      "epoch": 0.06513479283517279,
      "grad_norm": 0.7497789263725281,
      "learning_rate": 2.4353723016560527e-05,
      "loss": 1.5783,
      "step": 990
    },
    {
      "epoch": 0.06520058555520832,
      "grad_norm": 0.7946657538414001,
      "learning_rate": 2.4261864814450758e-05,
      "loss": 1.5937,
      "step": 991
    },
    {
      "epoch": 0.06526637827524384,
      "grad_norm": 0.7872624397277832,
      "learning_rate": 2.417012464886978e-05,
      "loss": 1.5904,
      "step": 992
    },
    {
      "epoch": 0.06533217099527937,
      "grad_norm": 0.8094744682312012,
      "learning_rate": 2.4078502940543157e-05,
      "loss": 1.5454,
      "step": 993
    },
    {
      "epoch": 0.0653979637153149,
      "grad_norm": 0.9983581304550171,
      "learning_rate": 2.3987000109653134e-05,
      "loss": 1.4996,
      "step": 994
    },
    {
      "epoch": 0.06546375643535043,
      "grad_norm": 0.826250433921814,
      "learning_rate": 2.389561657583681e-05,
      "loss": 1.5753,
      "step": 995
    },
    {
      "epoch": 0.06552954915538596,
      "grad_norm": 0.8517383337020874,
      "learning_rate": 2.3804352758184223e-05,
      "loss": 1.5688,
      "step": 996
    },
    {
      "epoch": 0.06559534187542149,
      "grad_norm": 0.8434426784515381,
      "learning_rate": 2.3713209075236343e-05,
      "loss": 1.3069,
      "step": 997
    },
    {
      "epoch": 0.06566113459545701,
      "grad_norm": 0.9146838784217834,
      "learning_rate": 2.3622185944983188e-05,
      "loss": 1.4057,
      "step": 998
    },
    {
      "epoch": 0.06572692731549254,
      "grad_norm": 1.0109113454818726,
      "learning_rate": 2.3531283784861952e-05,
      "loss": 1.6269,
      "step": 999
    },
    {
      "epoch": 0.06579272003552807,
      "grad_norm": 1.2136893272399902,
      "learning_rate": 2.3440503011755e-05,
      "loss": 1.5136,
      "step": 1000
    },
    {
      "epoch": 0.06585851275556359,
      "grad_norm": 0.5091771483421326,
      "learning_rate": 2.3349844041988045e-05,
      "loss": 1.7648,
      "step": 1001
    },
    {
      "epoch": 0.06592430547559912,
      "grad_norm": 0.5724332928657532,
      "learning_rate": 2.325930729132821e-05,
      "loss": 1.8152,
      "step": 1002
    },
    {
      "epoch": 0.06599009819563466,
      "grad_norm": 0.5698429942131042,
      "learning_rate": 2.3168893174982098e-05,
      "loss": 1.6787,
      "step": 1003
    },
    {
      "epoch": 0.06605589091567018,
      "grad_norm": 0.5834808945655823,
      "learning_rate": 2.30786021075939e-05,
      "loss": 1.6803,
      "step": 1004
    },
    {
      "epoch": 0.06612168363570571,
      "grad_norm": 0.588674008846283,
      "learning_rate": 2.2988434503243516e-05,
      "loss": 1.5799,
      "step": 1005
    },
    {
      "epoch": 0.06618747635574124,
      "grad_norm": 0.6044074296951294,
      "learning_rate": 2.289839077544463e-05,
      "loss": 1.7156,
      "step": 1006
    },
    {
      "epoch": 0.06625326907577676,
      "grad_norm": 0.6402976512908936,
      "learning_rate": 2.28084713371428e-05,
      "loss": 1.7041,
      "step": 1007
    },
    {
      "epoch": 0.06631906179581229,
      "grad_norm": 0.5715565085411072,
      "learning_rate": 2.2718676600713622e-05,
      "loss": 1.5912,
      "step": 1008
    },
    {
      "epoch": 0.06638485451584782,
      "grad_norm": 0.6274817585945129,
      "learning_rate": 2.2629006977960798e-05,
      "loss": 1.6121,
      "step": 1009
    },
    {
      "epoch": 0.06645064723588336,
      "grad_norm": 0.5526905655860901,
      "learning_rate": 2.2539462880114194e-05,
      "loss": 1.5085,
      "step": 1010
    },
    {
      "epoch": 0.06651643995591888,
      "grad_norm": 0.6097638010978699,
      "learning_rate": 2.245004471782806e-05,
      "loss": 1.5009,
      "step": 1011
    },
    {
      "epoch": 0.06658223267595441,
      "grad_norm": 0.6747389435768127,
      "learning_rate": 2.236075290117914e-05,
      "loss": 1.6578,
      "step": 1012
    },
    {
      "epoch": 0.06664802539598993,
      "grad_norm": 0.6369504928588867,
      "learning_rate": 2.2271587839664672e-05,
      "loss": 1.5528,
      "step": 1013
    },
    {
      "epoch": 0.06671381811602546,
      "grad_norm": 0.6106559038162231,
      "learning_rate": 2.218254994220062e-05,
      "loss": 1.5907,
      "step": 1014
    },
    {
      "epoch": 0.06677961083606099,
      "grad_norm": 0.5885758399963379,
      "learning_rate": 2.2093639617119794e-05,
      "loss": 1.513,
      "step": 1015
    },
    {
      "epoch": 0.06684540355609651,
      "grad_norm": 0.6991572380065918,
      "learning_rate": 2.2004857272169876e-05,
      "loss": 1.823,
      "step": 1016
    },
    {
      "epoch": 0.06691119627613204,
      "grad_norm": 0.6160642504692078,
      "learning_rate": 2.1916203314511692e-05,
      "loss": 1.6389,
      "step": 1017
    },
    {
      "epoch": 0.06697698899616758,
      "grad_norm": 0.6292829513549805,
      "learning_rate": 2.1827678150717256e-05,
      "loss": 1.6663,
      "step": 1018
    },
    {
      "epoch": 0.0670427817162031,
      "grad_norm": 0.6121276617050171,
      "learning_rate": 2.1739282186767923e-05,
      "loss": 1.5885,
      "step": 1019
    },
    {
      "epoch": 0.06710857443623863,
      "grad_norm": 0.6440593004226685,
      "learning_rate": 2.165101582805254e-05,
      "loss": 1.6319,
      "step": 1020
    },
    {
      "epoch": 0.06717436715627416,
      "grad_norm": 0.636544942855835,
      "learning_rate": 2.1562879479365556e-05,
      "loss": 1.644,
      "step": 1021
    },
    {
      "epoch": 0.06724015987630969,
      "grad_norm": 0.6443448662757874,
      "learning_rate": 2.1474873544905205e-05,
      "loss": 1.5995,
      "step": 1022
    },
    {
      "epoch": 0.06730595259634521,
      "grad_norm": 0.6514659523963928,
      "learning_rate": 2.1386998428271633e-05,
      "loss": 1.554,
      "step": 1023
    },
    {
      "epoch": 0.06737174531638074,
      "grad_norm": 0.6635521650314331,
      "learning_rate": 2.129925453246504e-05,
      "loss": 1.7461,
      "step": 1024
    },
    {
      "epoch": 0.06743753803641626,
      "grad_norm": 0.6211679577827454,
      "learning_rate": 2.1211642259883867e-05,
      "loss": 1.4678,
      "step": 1025
    },
    {
      "epoch": 0.0675033307564518,
      "grad_norm": 0.668057918548584,
      "learning_rate": 2.1124162012322862e-05,
      "loss": 1.6624,
      "step": 1026
    },
    {
      "epoch": 0.06756912347648733,
      "grad_norm": 0.6626843810081482,
      "learning_rate": 2.1036814190971333e-05,
      "loss": 1.5244,
      "step": 1027
    },
    {
      "epoch": 0.06763491619652286,
      "grad_norm": 0.6613166332244873,
      "learning_rate": 2.0949599196411325e-05,
      "loss": 1.497,
      "step": 1028
    },
    {
      "epoch": 0.06770070891655838,
      "grad_norm": 0.6597977876663208,
      "learning_rate": 2.086251742861565e-05,
      "loss": 1.711,
      "step": 1029
    },
    {
      "epoch": 0.06776650163659391,
      "grad_norm": 0.7360920310020447,
      "learning_rate": 2.077556928694619e-05,
      "loss": 1.6281,
      "step": 1030
    },
    {
      "epoch": 0.06783229435662944,
      "grad_norm": 0.6604809761047363,
      "learning_rate": 2.0688755170151996e-05,
      "loss": 1.6061,
      "step": 1031
    },
    {
      "epoch": 0.06789808707666496,
      "grad_norm": 0.6945058703422546,
      "learning_rate": 2.0602075476367432e-05,
      "loss": 1.6064,
      "step": 1032
    },
    {
      "epoch": 0.06796387979670049,
      "grad_norm": 0.71868896484375,
      "learning_rate": 2.051553060311045e-05,
      "loss": 1.56,
      "step": 1033
    },
    {
      "epoch": 0.06802967251673603,
      "grad_norm": 0.6665169596672058,
      "learning_rate": 2.0429120947280678e-05,
      "loss": 1.5953,
      "step": 1034
    },
    {
      "epoch": 0.06809546523677155,
      "grad_norm": 0.6772186756134033,
      "learning_rate": 2.0342846905157636e-05,
      "loss": 1.5658,
      "step": 1035
    },
    {
      "epoch": 0.06816125795680708,
      "grad_norm": 0.7677963376045227,
      "learning_rate": 2.0256708872398915e-05,
      "loss": 1.7299,
      "step": 1036
    },
    {
      "epoch": 0.06822705067684261,
      "grad_norm": 0.7391518950462341,
      "learning_rate": 2.017070724403835e-05,
      "loss": 1.6122,
      "step": 1037
    },
    {
      "epoch": 0.06829284339687813,
      "grad_norm": 0.8133260607719421,
      "learning_rate": 2.0084842414484222e-05,
      "loss": 1.5689,
      "step": 1038
    },
    {
      "epoch": 0.06835863611691366,
      "grad_norm": 0.7881767749786377,
      "learning_rate": 1.999911477751746e-05,
      "loss": 1.673,
      "step": 1039
    },
    {
      "epoch": 0.06842442883694919,
      "grad_norm": 0.7791821360588074,
      "learning_rate": 1.9913524726289784e-05,
      "loss": 1.7744,
      "step": 1040
    },
    {
      "epoch": 0.06849022155698473,
      "grad_norm": 0.7703235745429993,
      "learning_rate": 1.9828072653321978e-05,
      "loss": 1.4325,
      "step": 1041
    },
    {
      "epoch": 0.06855601427702025,
      "grad_norm": 0.7927573323249817,
      "learning_rate": 1.9742758950502045e-05,
      "loss": 1.6792,
      "step": 1042
    },
    {
      "epoch": 0.06862180699705578,
      "grad_norm": 0.7753776907920837,
      "learning_rate": 1.965758400908334e-05,
      "loss": 1.4926,
      "step": 1043
    },
    {
      "epoch": 0.0686875997170913,
      "grad_norm": 0.8299908638000488,
      "learning_rate": 1.957254821968298e-05,
      "loss": 1.743,
      "step": 1044
    },
    {
      "epoch": 0.06875339243712683,
      "grad_norm": 0.754217267036438,
      "learning_rate": 1.948765197227983e-05,
      "loss": 1.4112,
      "step": 1045
    },
    {
      "epoch": 0.06881918515716236,
      "grad_norm": 0.7843993902206421,
      "learning_rate": 1.9402895656212833e-05,
      "loss": 1.4893,
      "step": 1046
    },
    {
      "epoch": 0.06888497787719788,
      "grad_norm": 0.8628965616226196,
      "learning_rate": 1.9318279660179217e-05,
      "loss": 1.5237,
      "step": 1047
    },
    {
      "epoch": 0.06895077059723341,
      "grad_norm": 0.9835182428359985,
      "learning_rate": 1.9233804372232643e-05,
      "loss": 1.5732,
      "step": 1048
    },
    {
      "epoch": 0.06901656331726895,
      "grad_norm": 1.099034309387207,
      "learning_rate": 1.914947017978153e-05,
      "loss": 1.5774,
      "step": 1049
    },
    {
      "epoch": 0.06908235603730448,
      "grad_norm": 1.265041470527649,
      "learning_rate": 1.9065277469587204e-05,
      "loss": 1.3189,
      "step": 1050
    },
    {
      "epoch": 0.06914814875734,
      "grad_norm": 0.5150356888771057,
      "learning_rate": 1.8981226627762143e-05,
      "loss": 1.8932,
      "step": 1051
    },
    {
      "epoch": 0.06921394147737553,
      "grad_norm": 0.5652185678482056,
      "learning_rate": 1.889731803976822e-05,
      "loss": 1.8346,
      "step": 1052
    },
    {
      "epoch": 0.06927973419741106,
      "grad_norm": 0.5585607886314392,
      "learning_rate": 1.881355209041491e-05,
      "loss": 1.7214,
      "step": 1053
    },
    {
      "epoch": 0.06934552691744658,
      "grad_norm": 0.5780192613601685,
      "learning_rate": 1.8729929163857552e-05,
      "loss": 1.6242,
      "step": 1054
    },
    {
      "epoch": 0.06941131963748211,
      "grad_norm": 0.5689457654953003,
      "learning_rate": 1.8646449643595565e-05,
      "loss": 1.6942,
      "step": 1055
    },
    {
      "epoch": 0.06947711235751763,
      "grad_norm": 0.5897836685180664,
      "learning_rate": 1.8563113912470702e-05,
      "loss": 1.6164,
      "step": 1056
    },
    {
      "epoch": 0.06954290507755317,
      "grad_norm": 0.5599413514137268,
      "learning_rate": 1.847992235266529e-05,
      "loss": 1.5614,
      "step": 1057
    },
    {
      "epoch": 0.0696086977975887,
      "grad_norm": 0.5642629861831665,
      "learning_rate": 1.8396875345700497e-05,
      "loss": 1.54,
      "step": 1058
    },
    {
      "epoch": 0.06967449051762423,
      "grad_norm": 0.6085110902786255,
      "learning_rate": 1.8313973272434504e-05,
      "loss": 1.6037,
      "step": 1059
    },
    {
      "epoch": 0.06974028323765975,
      "grad_norm": 0.5913036465644836,
      "learning_rate": 1.8231216513060893e-05,
      "loss": 1.6223,
      "step": 1060
    },
    {
      "epoch": 0.06980607595769528,
      "grad_norm": 0.5918750166893005,
      "learning_rate": 1.8148605447106797e-05,
      "loss": 1.5262,
      "step": 1061
    },
    {
      "epoch": 0.0698718686777308,
      "grad_norm": 0.5838180184364319,
      "learning_rate": 1.8066140453431173e-05,
      "loss": 1.6248,
      "step": 1062
    },
    {
      "epoch": 0.06993766139776633,
      "grad_norm": 0.6557063460350037,
      "learning_rate": 1.798382191022313e-05,
      "loss": 1.7616,
      "step": 1063
    },
    {
      "epoch": 0.07000345411780187,
      "grad_norm": 0.6090285181999207,
      "learning_rate": 1.7901650195000068e-05,
      "loss": 1.6591,
      "step": 1064
    },
    {
      "epoch": 0.0700692468378374,
      "grad_norm": 0.6101527214050293,
      "learning_rate": 1.7819625684606102e-05,
      "loss": 1.5755,
      "step": 1065
    },
    {
      "epoch": 0.07013503955787292,
      "grad_norm": 0.5960958003997803,
      "learning_rate": 1.7737748755210214e-05,
      "loss": 1.5059,
      "step": 1066
    },
    {
      "epoch": 0.07020083227790845,
      "grad_norm": 0.631040096282959,
      "learning_rate": 1.76560197823046e-05,
      "loss": 1.5921,
      "step": 1067
    },
    {
      "epoch": 0.07026662499794398,
      "grad_norm": 0.6434530019760132,
      "learning_rate": 1.7574439140702902e-05,
      "loss": 1.5911,
      "step": 1068
    },
    {
      "epoch": 0.0703324177179795,
      "grad_norm": 0.6554668545722961,
      "learning_rate": 1.7493007204538503e-05,
      "loss": 1.7435,
      "step": 1069
    },
    {
      "epoch": 0.07039821043801503,
      "grad_norm": 0.6489726901054382,
      "learning_rate": 1.7411724347262824e-05,
      "loss": 1.7055,
      "step": 1070
    },
    {
      "epoch": 0.07046400315805056,
      "grad_norm": 0.628710150718689,
      "learning_rate": 1.7330590941643603e-05,
      "loss": 1.5092,
      "step": 1071
    },
    {
      "epoch": 0.0705297958780861,
      "grad_norm": 0.6012537479400635,
      "learning_rate": 1.7249607359763174e-05,
      "loss": 1.4466,
      "step": 1072
    },
    {
      "epoch": 0.07059558859812162,
      "grad_norm": 0.6696065664291382,
      "learning_rate": 1.7168773973016776e-05,
      "loss": 1.5727,
      "step": 1073
    },
    {
      "epoch": 0.07066138131815715,
      "grad_norm": 0.6296167373657227,
      "learning_rate": 1.7088091152110873e-05,
      "loss": 1.5393,
      "step": 1074
    },
    {
      "epoch": 0.07072717403819268,
      "grad_norm": 0.6769093871116638,
      "learning_rate": 1.7007559267061334e-05,
      "loss": 1.6523,
      "step": 1075
    },
    {
      "epoch": 0.0707929667582282,
      "grad_norm": 0.6519213914871216,
      "learning_rate": 1.692717868719195e-05,
      "loss": 1.5566,
      "step": 1076
    },
    {
      "epoch": 0.07085875947826373,
      "grad_norm": 0.6642208695411682,
      "learning_rate": 1.6846949781132548e-05,
      "loss": 1.5886,
      "step": 1077
    },
    {
      "epoch": 0.07092455219829925,
      "grad_norm": 0.6803880333900452,
      "learning_rate": 1.6766872916817377e-05,
      "loss": 1.6519,
      "step": 1078
    },
    {
      "epoch": 0.07099034491833478,
      "grad_norm": 0.7287022471427917,
      "learning_rate": 1.668694846148343e-05,
      "loss": 1.7061,
      "step": 1079
    },
    {
      "epoch": 0.07105613763837032,
      "grad_norm": 0.6571015119552612,
      "learning_rate": 1.660717678166871e-05,
      "loss": 1.5716,
      "step": 1080
    },
    {
      "epoch": 0.07112193035840585,
      "grad_norm": 0.6374773383140564,
      "learning_rate": 1.6527558243210623e-05,
      "loss": 1.455,
      "step": 1081
    },
    {
      "epoch": 0.07118772307844137,
      "grad_norm": 0.699365496635437,
      "learning_rate": 1.644809321124423e-05,
      "loss": 1.6404,
      "step": 1082
    },
    {
      "epoch": 0.0712535157984769,
      "grad_norm": 0.6866129636764526,
      "learning_rate": 1.6368782050200633e-05,
      "loss": 1.4377,
      "step": 1083
    },
    {
      "epoch": 0.07131930851851243,
      "grad_norm": 0.6566864252090454,
      "learning_rate": 1.6289625123805245e-05,
      "loss": 1.481,
      "step": 1084
    },
    {
      "epoch": 0.07138510123854795,
      "grad_norm": 0.7273065447807312,
      "learning_rate": 1.621062279507617e-05,
      "loss": 1.6732,
      "step": 1085
    },
    {
      "epoch": 0.07145089395858348,
      "grad_norm": 0.7388779520988464,
      "learning_rate": 1.6131775426322503e-05,
      "loss": 1.5516,
      "step": 1086
    },
    {
      "epoch": 0.071516686678619,
      "grad_norm": 0.7159716486930847,
      "learning_rate": 1.60530833791427e-05,
      "loss": 1.5815,
      "step": 1087
    },
    {
      "epoch": 0.07158247939865454,
      "grad_norm": 0.7883169054985046,
      "learning_rate": 1.597454701442288e-05,
      "loss": 1.5699,
      "step": 1088
    },
    {
      "epoch": 0.07164827211869007,
      "grad_norm": 0.7535433769226074,
      "learning_rate": 1.589616669233522e-05,
      "loss": 1.71,
      "step": 1089
    },
    {
      "epoch": 0.0717140648387256,
      "grad_norm": 0.7542433142662048,
      "learning_rate": 1.581794277233628e-05,
      "loss": 1.6412,
      "step": 1090
    },
    {
      "epoch": 0.07177985755876112,
      "grad_norm": 0.7680163383483887,
      "learning_rate": 1.5739875613165283e-05,
      "loss": 1.5868,
      "step": 1091
    },
    {
      "epoch": 0.07184565027879665,
      "grad_norm": 0.8039641380310059,
      "learning_rate": 1.566196557284264e-05,
      "loss": 1.6165,
      "step": 1092
    },
    {
      "epoch": 0.07191144299883218,
      "grad_norm": 0.8360242247581482,
      "learning_rate": 1.5584213008668147e-05,
      "loss": 1.4836,
      "step": 1093
    },
    {
      "epoch": 0.0719772357188677,
      "grad_norm": 0.9569554328918457,
      "learning_rate": 1.550661827721941e-05,
      "loss": 1.7255,
      "step": 1094
    },
    {
      "epoch": 0.07204302843890324,
      "grad_norm": 0.8485451340675354,
      "learning_rate": 1.5429181734350236e-05,
      "loss": 1.6198,
      "step": 1095
    },
    {
      "epoch": 0.07210882115893877,
      "grad_norm": 0.8728616237640381,
      "learning_rate": 1.53519037351889e-05,
      "loss": 1.6094,
      "step": 1096
    },
    {
      "epoch": 0.0721746138789743,
      "grad_norm": 0.9522159695625305,
      "learning_rate": 1.527478463413666e-05,
      "loss": 1.3929,
      "step": 1097
    },
    {
      "epoch": 0.07224040659900982,
      "grad_norm": 1.0558445453643799,
      "learning_rate": 1.5197824784866015e-05,
      "loss": 1.7511,
      "step": 1098
    },
    {
      "epoch": 0.07230619931904535,
      "grad_norm": 1.0129352807998657,
      "learning_rate": 1.5121024540319161e-05,
      "loss": 1.4986,
      "step": 1099
    },
    {
      "epoch": 0.07237199203908087,
      "grad_norm": 1.232009768486023,
      "learning_rate": 1.5044384252706312e-05,
      "loss": 1.2485,
      "step": 1100
    },
    {
      "epoch": 0.0724377847591164,
      "grad_norm": 0.5130254626274109,
      "learning_rate": 1.4967904273504113e-05,
      "loss": 1.816,
      "step": 1101
    },
    {
      "epoch": 0.07250357747915193,
      "grad_norm": 0.5423897504806519,
      "learning_rate": 1.4891584953454036e-05,
      "loss": 1.7201,
      "step": 1102
    },
    {
      "epoch": 0.07256937019918747,
      "grad_norm": 0.5203692317008972,
      "learning_rate": 1.4815426642560754e-05,
      "loss": 1.5166,
      "step": 1103
    },
    {
      "epoch": 0.07263516291922299,
      "grad_norm": 0.5805792212486267,
      "learning_rate": 1.4739429690090533e-05,
      "loss": 1.6115,
      "step": 1104
    },
    {
      "epoch": 0.07270095563925852,
      "grad_norm": 0.5422717928886414,
      "learning_rate": 1.4663594444569667e-05,
      "loss": 1.5951,
      "step": 1105
    },
    {
      "epoch": 0.07276674835929405,
      "grad_norm": 0.5676501989364624,
      "learning_rate": 1.4587921253782849e-05,
      "loss": 1.6266,
      "step": 1106
    },
    {
      "epoch": 0.07283254107932957,
      "grad_norm": 0.5662168264389038,
      "learning_rate": 1.4512410464771514e-05,
      "loss": 1.6486,
      "step": 1107
    },
    {
      "epoch": 0.0728983337993651,
      "grad_norm": 0.5962639451026917,
      "learning_rate": 1.4437062423832426e-05,
      "loss": 1.8598,
      "step": 1108
    },
    {
      "epoch": 0.07296412651940062,
      "grad_norm": 0.6214326620101929,
      "learning_rate": 1.4361877476515889e-05,
      "loss": 1.5881,
      "step": 1109
    },
    {
      "epoch": 0.07302991923943615,
      "grad_norm": 0.5972443222999573,
      "learning_rate": 1.428685596762429e-05,
      "loss": 1.6234,
      "step": 1110
    },
    {
      "epoch": 0.07309571195947169,
      "grad_norm": 0.5804590582847595,
      "learning_rate": 1.4211998241210484e-05,
      "loss": 1.7503,
      "step": 1111
    },
    {
      "epoch": 0.07316150467950722,
      "grad_norm": 0.6213569045066833,
      "learning_rate": 1.413730464057616e-05,
      "loss": 1.6576,
      "step": 1112
    },
    {
      "epoch": 0.07322729739954274,
      "grad_norm": 0.6171157360076904,
      "learning_rate": 1.406277550827037e-05,
      "loss": 1.6889,
      "step": 1113
    },
    {
      "epoch": 0.07329309011957827,
      "grad_norm": 0.63736891746521,
      "learning_rate": 1.3988411186087885e-05,
      "loss": 1.5542,
      "step": 1114
    },
    {
      "epoch": 0.0733588828396138,
      "grad_norm": 0.6111243367195129,
      "learning_rate": 1.3914212015067651e-05,
      "loss": 1.6043,
      "step": 1115
    },
    {
      "epoch": 0.07342467555964932,
      "grad_norm": 0.6354307532310486,
      "learning_rate": 1.3840178335491222e-05,
      "loss": 1.5426,
      "step": 1116
    },
    {
      "epoch": 0.07349046827968485,
      "grad_norm": 0.6276394128799438,
      "learning_rate": 1.376631048688119e-05,
      "loss": 1.5725,
      "step": 1117
    },
    {
      "epoch": 0.07355626099972037,
      "grad_norm": 0.6149314641952515,
      "learning_rate": 1.3692608807999652e-05,
      "loss": 1.6483,
      "step": 1118
    },
    {
      "epoch": 0.07362205371975591,
      "grad_norm": 0.6823477745056152,
      "learning_rate": 1.3619073636846625e-05,
      "loss": 1.6028,
      "step": 1119
    },
    {
      "epoch": 0.07368784643979144,
      "grad_norm": 0.6506856083869934,
      "learning_rate": 1.3545705310658529e-05,
      "loss": 1.5915,
      "step": 1120
    },
    {
      "epoch": 0.07375363915982697,
      "grad_norm": 0.6432639956474304,
      "learning_rate": 1.3472504165906613e-05,
      "loss": 1.5872,
      "step": 1121
    },
    {
      "epoch": 0.0738194318798625,
      "grad_norm": 0.6213419437408447,
      "learning_rate": 1.3399470538295434e-05,
      "loss": 1.5422,
      "step": 1122
    },
    {
      "epoch": 0.07388522459989802,
      "grad_norm": 0.6513727903366089,
      "learning_rate": 1.3326604762761258e-05,
      "loss": 1.6088,
      "step": 1123
    },
    {
      "epoch": 0.07395101731993355,
      "grad_norm": 0.6472317576408386,
      "learning_rate": 1.3253907173470648e-05,
      "loss": 1.5638,
      "step": 1124
    },
    {
      "epoch": 0.07401681003996907,
      "grad_norm": 0.6375536322593689,
      "learning_rate": 1.3181378103818814e-05,
      "loss": 1.4387,
      "step": 1125
    },
    {
      "epoch": 0.07408260276000461,
      "grad_norm": 0.6542819738388062,
      "learning_rate": 1.3109017886428122e-05,
      "loss": 1.5019,
      "step": 1126
    },
    {
      "epoch": 0.07414839548004014,
      "grad_norm": 0.6725215911865234,
      "learning_rate": 1.30368268531466e-05,
      "loss": 1.5493,
      "step": 1127
    },
    {
      "epoch": 0.07421418820007566,
      "grad_norm": 0.695514976978302,
      "learning_rate": 1.2964805335046332e-05,
      "loss": 1.7503,
      "step": 1128
    },
    {
      "epoch": 0.07427998092011119,
      "grad_norm": 0.6565685868263245,
      "learning_rate": 1.2892953662422047e-05,
      "loss": 1.4902,
      "step": 1129
    },
    {
      "epoch": 0.07434577364014672,
      "grad_norm": 0.7066967487335205,
      "learning_rate": 1.2821272164789544e-05,
      "loss": 1.5197,
      "step": 1130
    },
    {
      "epoch": 0.07441156636018224,
      "grad_norm": 0.686930239200592,
      "learning_rate": 1.2749761170884179e-05,
      "loss": 1.596,
      "step": 1131
    },
    {
      "epoch": 0.07447735908021777,
      "grad_norm": 0.7322065830230713,
      "learning_rate": 1.2678421008659375e-05,
      "loss": 1.5857,
      "step": 1132
    },
    {
      "epoch": 0.0745431518002533,
      "grad_norm": 0.6980546116828918,
      "learning_rate": 1.2607252005285109e-05,
      "loss": 1.5733,
      "step": 1133
    },
    {
      "epoch": 0.07460894452028884,
      "grad_norm": 0.7424272298812866,
      "learning_rate": 1.2536254487146415e-05,
      "loss": 1.6048,
      "step": 1134
    },
    {
      "epoch": 0.07467473724032436,
      "grad_norm": 0.708510160446167,
      "learning_rate": 1.2465428779841882e-05,
      "loss": 1.5323,
      "step": 1135
    },
    {
      "epoch": 0.07474052996035989,
      "grad_norm": 0.7060134410858154,
      "learning_rate": 1.2394775208182174e-05,
      "loss": 1.5616,
      "step": 1136
    },
    {
      "epoch": 0.07480632268039542,
      "grad_norm": 0.7687208652496338,
      "learning_rate": 1.2324294096188526e-05,
      "loss": 1.6314,
      "step": 1137
    },
    {
      "epoch": 0.07487211540043094,
      "grad_norm": 0.7260802388191223,
      "learning_rate": 1.2253985767091274e-05,
      "loss": 1.5503,
      "step": 1138
    },
    {
      "epoch": 0.07493790812046647,
      "grad_norm": 0.7328141331672668,
      "learning_rate": 1.2183850543328312e-05,
      "loss": 1.5422,
      "step": 1139
    },
    {
      "epoch": 0.075003700840502,
      "grad_norm": 0.7931233048439026,
      "learning_rate": 1.2113888746543738e-05,
      "loss": 1.4909,
      "step": 1140
    },
    {
      "epoch": 0.07506949356053752,
      "grad_norm": 0.7329046130180359,
      "learning_rate": 1.2044100697586263e-05,
      "loss": 1.4206,
      "step": 1141
    },
    {
      "epoch": 0.07513528628057306,
      "grad_norm": 0.8105408549308777,
      "learning_rate": 1.1974486716507783e-05,
      "loss": 1.6199,
      "step": 1142
    },
    {
      "epoch": 0.07520107900060859,
      "grad_norm": 0.8885856866836548,
      "learning_rate": 1.1905047122561924e-05,
      "loss": 1.7629,
      "step": 1143
    },
    {
      "epoch": 0.07526687172064411,
      "grad_norm": 0.8523672223091125,
      "learning_rate": 1.1835782234202525e-05,
      "loss": 1.505,
      "step": 1144
    },
    {
      "epoch": 0.07533266444067964,
      "grad_norm": 0.8462732434272766,
      "learning_rate": 1.1766692369082255e-05,
      "loss": 1.5047,
      "step": 1145
    },
    {
      "epoch": 0.07539845716071517,
      "grad_norm": 0.8410788178443909,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.5502,
      "step": 1146
    },
    {
      "epoch": 0.07546424988075069,
      "grad_norm": 0.8786489367485046,
      "learning_rate": 1.1629038975154943e-05,
      "loss": 1.4928,
      "step": 1147
    },
    {
      "epoch": 0.07553004260078622,
      "grad_norm": 0.9621903896331787,
      "learning_rate": 1.156047607763407e-05,
      "loss": 1.4004,
      "step": 1148
    },
    {
      "epoch": 0.07559583532082174,
      "grad_norm": 1.068969488143921,
      "learning_rate": 1.1492089465921768e-05,
      "loss": 1.5004,
      "step": 1149
    },
    {
      "epoch": 0.07566162804085728,
      "grad_norm": 1.3039774894714355,
      "learning_rate": 1.1423879453642878e-05,
      "loss": 1.5061,
      "step": 1150
    },
    {
      "epoch": 0.07572742076089281,
      "grad_norm": 0.5217190384864807,
      "learning_rate": 1.135584635361232e-05,
      "loss": 1.8168,
      "step": 1151
    },
    {
      "epoch": 0.07579321348092834,
      "grad_norm": 0.5663981437683105,
      "learning_rate": 1.128799047783371e-05,
      "loss": 1.7287,
      "step": 1152
    },
    {
      "epoch": 0.07585900620096386,
      "grad_norm": 0.5568148493766785,
      "learning_rate": 1.122031213749789e-05,
      "loss": 1.611,
      "step": 1153
    },
    {
      "epoch": 0.07592479892099939,
      "grad_norm": 0.6089836359024048,
      "learning_rate": 1.115281164298153e-05,
      "loss": 1.5694,
      "step": 1154
    },
    {
      "epoch": 0.07599059164103492,
      "grad_norm": 0.5871589183807373,
      "learning_rate": 1.1085489303845637e-05,
      "loss": 1.6517,
      "step": 1155
    },
    {
      "epoch": 0.07605638436107044,
      "grad_norm": 0.5930696725845337,
      "learning_rate": 1.101834542883427e-05,
      "loss": 1.6933,
      "step": 1156
    },
    {
      "epoch": 0.07612217708110598,
      "grad_norm": 0.5683667659759521,
      "learning_rate": 1.0951380325872979e-05,
      "loss": 1.6172,
      "step": 1157
    },
    {
      "epoch": 0.07618796980114151,
      "grad_norm": 0.5930812954902649,
      "learning_rate": 1.088459430206748e-05,
      "loss": 1.7436,
      "step": 1158
    },
    {
      "epoch": 0.07625376252117703,
      "grad_norm": 0.5984984040260315,
      "learning_rate": 1.0817987663702229e-05,
      "loss": 1.8179,
      "step": 1159
    },
    {
      "epoch": 0.07631955524121256,
      "grad_norm": 0.6047726273536682,
      "learning_rate": 1.0751560716238967e-05,
      "loss": 1.5642,
      "step": 1160
    },
    {
      "epoch": 0.07638534796124809,
      "grad_norm": 0.6473960280418396,
      "learning_rate": 1.0685313764315413e-05,
      "loss": 1.7192,
      "step": 1161
    },
    {
      "epoch": 0.07645114068128361,
      "grad_norm": 0.6276887059211731,
      "learning_rate": 1.0619247111743797e-05,
      "loss": 1.532,
      "step": 1162
    },
    {
      "epoch": 0.07651693340131914,
      "grad_norm": 0.6259212493896484,
      "learning_rate": 1.055336106150948e-05,
      "loss": 1.5419,
      "step": 1163
    },
    {
      "epoch": 0.07658272612135467,
      "grad_norm": 0.6881235241889954,
      "learning_rate": 1.048765591576959e-05,
      "loss": 1.7839,
      "step": 1164
    },
    {
      "epoch": 0.0766485188413902,
      "grad_norm": 0.6311882138252258,
      "learning_rate": 1.0422131975851584e-05,
      "loss": 1.5377,
      "step": 1165
    },
    {
      "epoch": 0.07671431156142573,
      "grad_norm": 0.6650845408439636,
      "learning_rate": 1.0356789542251938e-05,
      "loss": 1.7409,
      "step": 1166
    },
    {
      "epoch": 0.07678010428146126,
      "grad_norm": 0.6411380767822266,
      "learning_rate": 1.0291628914634694e-05,
      "loss": 1.6915,
      "step": 1167
    },
    {
      "epoch": 0.07684589700149679,
      "grad_norm": 0.6582992672920227,
      "learning_rate": 1.022665039183015e-05,
      "loss": 1.5401,
      "step": 1168
    },
    {
      "epoch": 0.07691168972153231,
      "grad_norm": 0.6362737417221069,
      "learning_rate": 1.0161854271833443e-05,
      "loss": 1.5978,
      "step": 1169
    },
    {
      "epoch": 0.07697748244156784,
      "grad_norm": 0.6066959500312805,
      "learning_rate": 1.009724085180322e-05,
      "loss": 1.6114,
      "step": 1170
    },
    {
      "epoch": 0.07704327516160336,
      "grad_norm": 0.6722879409790039,
      "learning_rate": 1.0032810428060218e-05,
      "loss": 1.597,
      "step": 1171
    },
    {
      "epoch": 0.07710906788163889,
      "grad_norm": 0.6590387225151062,
      "learning_rate": 9.96856329608597e-06,
      "loss": 1.5352,
      "step": 1172
    },
    {
      "epoch": 0.07717486060167443,
      "grad_norm": 0.6803810000419617,
      "learning_rate": 9.90449975052144e-06,
      "loss": 1.5224,
      "step": 1173
    },
    {
      "epoch": 0.07724065332170996,
      "grad_norm": 0.6694747805595398,
      "learning_rate": 9.840620085165626e-06,
      "loss": 1.7062,
      "step": 1174
    },
    {
      "epoch": 0.07730644604174548,
      "grad_norm": 0.672380268573761,
      "learning_rate": 9.776924592974256e-06,
      "loss": 1.6693,
      "step": 1175
    },
    {
      "epoch": 0.07737223876178101,
      "grad_norm": 0.6489235162734985,
      "learning_rate": 9.713413566058405e-06,
      "loss": 1.5843,
      "step": 1176
    },
    {
      "epoch": 0.07743803148181654,
      "grad_norm": 0.6905193328857422,
      "learning_rate": 9.650087295683202e-06,
      "loss": 1.593,
      "step": 1177
    },
    {
      "epoch": 0.07750382420185206,
      "grad_norm": 0.7134239673614502,
      "learning_rate": 9.586946072266478e-06,
      "loss": 1.5109,
      "step": 1178
    },
    {
      "epoch": 0.07756961692188759,
      "grad_norm": 0.6946320533752441,
      "learning_rate": 9.523990185377413e-06,
      "loss": 1.5549,
      "step": 1179
    },
    {
      "epoch": 0.07763540964192311,
      "grad_norm": 0.6910642981529236,
      "learning_rate": 9.461219923735227e-06,
      "loss": 1.7573,
      "step": 1180
    },
    {
      "epoch": 0.07770120236195865,
      "grad_norm": 0.6611814498901367,
      "learning_rate": 9.398635575207854e-06,
      "loss": 1.5795,
      "step": 1181
    },
    {
      "epoch": 0.07776699508199418,
      "grad_norm": 0.6925032734870911,
      "learning_rate": 9.336237426810624e-06,
      "loss": 1.6087,
      "step": 1182
    },
    {
      "epoch": 0.07783278780202971,
      "grad_norm": 0.7224239706993103,
      "learning_rate": 9.274025764704936e-06,
      "loss": 1.5171,
      "step": 1183
    },
    {
      "epoch": 0.07789858052206523,
      "grad_norm": 0.696713387966156,
      "learning_rate": 9.212000874196953e-06,
      "loss": 1.4585,
      "step": 1184
    },
    {
      "epoch": 0.07796437324210076,
      "grad_norm": 0.7397194504737854,
      "learning_rate": 9.150163039736297e-06,
      "loss": 1.551,
      "step": 1185
    },
    {
      "epoch": 0.07803016596213629,
      "grad_norm": 0.696258008480072,
      "learning_rate": 9.08851254491475e-06,
      "loss": 1.5841,
      "step": 1186
    },
    {
      "epoch": 0.07809595868217181,
      "grad_norm": 0.7256150841712952,
      "learning_rate": 9.027049672464916e-06,
      "loss": 1.6002,
      "step": 1187
    },
    {
      "epoch": 0.07816175140220735,
      "grad_norm": 0.7567223906517029,
      "learning_rate": 8.965774704258956e-06,
      "loss": 1.6436,
      "step": 1188
    },
    {
      "epoch": 0.07822754412224288,
      "grad_norm": 0.7757462859153748,
      "learning_rate": 8.90468792130733e-06,
      "loss": 1.7119,
      "step": 1189
    },
    {
      "epoch": 0.0782933368422784,
      "grad_norm": 0.7670205235481262,
      "learning_rate": 8.843789603757446e-06,
      "loss": 1.6326,
      "step": 1190
    },
    {
      "epoch": 0.07835912956231393,
      "grad_norm": 0.8732237219810486,
      "learning_rate": 8.783080030892394e-06,
      "loss": 1.415,
      "step": 1191
    },
    {
      "epoch": 0.07842492228234946,
      "grad_norm": 0.8682703375816345,
      "learning_rate": 8.72255948112966e-06,
      "loss": 1.629,
      "step": 1192
    },
    {
      "epoch": 0.07849071500238498,
      "grad_norm": 0.7908897399902344,
      "learning_rate": 8.662228232019876e-06,
      "loss": 1.3307,
      "step": 1193
    },
    {
      "epoch": 0.07855650772242051,
      "grad_norm": 0.8436971306800842,
      "learning_rate": 8.602086560245537e-06,
      "loss": 1.5682,
      "step": 1194
    },
    {
      "epoch": 0.07862230044245604,
      "grad_norm": 0.8712666630744934,
      "learning_rate": 8.542134741619711e-06,
      "loss": 1.5686,
      "step": 1195
    },
    {
      "epoch": 0.07868809316249158,
      "grad_norm": 0.8938828110694885,
      "learning_rate": 8.48237305108479e-06,
      "loss": 1.5938,
      "step": 1196
    },
    {
      "epoch": 0.0787538858825271,
      "grad_norm": 0.8834605813026428,
      "learning_rate": 8.422801762711247e-06,
      "loss": 1.4887,
      "step": 1197
    },
    {
      "epoch": 0.07881967860256263,
      "grad_norm": 1.0207496881484985,
      "learning_rate": 8.363421149696332e-06,
      "loss": 1.576,
      "step": 1198
    },
    {
      "epoch": 0.07888547132259816,
      "grad_norm": 1.0710493326187134,
      "learning_rate": 8.304231484362868e-06,
      "loss": 1.6295,
      "step": 1199
    },
    {
      "epoch": 0.07895126404263368,
      "grad_norm": 1.221801996231079,
      "learning_rate": 8.245233038157962e-06,
      "loss": 1.6023,
      "step": 1200
    },
    {
      "epoch": 0.07901705676266921,
      "grad_norm": 0.517112672328949,
      "learning_rate": 8.186426081651804e-06,
      "loss": 1.6261,
      "step": 1201
    },
    {
      "epoch": 0.07908284948270473,
      "grad_norm": 0.5203869938850403,
      "learning_rate": 8.127810884536403e-06,
      "loss": 1.6803,
      "step": 1202
    },
    {
      "epoch": 0.07914864220274026,
      "grad_norm": 0.526978611946106,
      "learning_rate": 8.069387715624294e-06,
      "loss": 1.6658,
      "step": 1203
    },
    {
      "epoch": 0.0792144349227758,
      "grad_norm": 0.5750939249992371,
      "learning_rate": 8.011156842847412e-06,
      "loss": 1.6741,
      "step": 1204
    },
    {
      "epoch": 0.07928022764281133,
      "grad_norm": 0.5610478520393372,
      "learning_rate": 7.95311853325582e-06,
      "loss": 1.6639,
      "step": 1205
    },
    {
      "epoch": 0.07934602036284685,
      "grad_norm": 0.5532097816467285,
      "learning_rate": 7.89527305301645e-06,
      "loss": 1.6807,
      "step": 1206
    },
    {
      "epoch": 0.07941181308288238,
      "grad_norm": 0.5609330534934998,
      "learning_rate": 7.83762066741191e-06,
      "loss": 1.701,
      "step": 1207
    },
    {
      "epoch": 0.0794776058029179,
      "grad_norm": 0.5949382781982422,
      "learning_rate": 7.780161640839257e-06,
      "loss": 1.7165,
      "step": 1208
    },
    {
      "epoch": 0.07954339852295343,
      "grad_norm": 0.5886333584785461,
      "learning_rate": 7.722896236808807e-06,
      "loss": 1.5563,
      "step": 1209
    },
    {
      "epoch": 0.07960919124298896,
      "grad_norm": 0.6132276654243469,
      "learning_rate": 7.665824717942915e-06,
      "loss": 1.7714,
      "step": 1210
    },
    {
      "epoch": 0.0796749839630245,
      "grad_norm": 0.6169953942298889,
      "learning_rate": 7.60894734597476e-06,
      "loss": 1.6546,
      "step": 1211
    },
    {
      "epoch": 0.07974077668306002,
      "grad_norm": 0.6119076609611511,
      "learning_rate": 7.552264381747148e-06,
      "loss": 1.6585,
      "step": 1212
    },
    {
      "epoch": 0.07980656940309555,
      "grad_norm": 0.6173310279846191,
      "learning_rate": 7.495776085211331e-06,
      "loss": 1.4996,
      "step": 1213
    },
    {
      "epoch": 0.07987236212313108,
      "grad_norm": 0.590090811252594,
      "learning_rate": 7.439482715425805e-06,
      "loss": 1.5205,
      "step": 1214
    },
    {
      "epoch": 0.0799381548431666,
      "grad_norm": 0.6101630926132202,
      "learning_rate": 7.383384530555104e-06,
      "loss": 1.681,
      "step": 1215
    },
    {
      "epoch": 0.08000394756320213,
      "grad_norm": 0.6025815010070801,
      "learning_rate": 7.327481787868646e-06,
      "loss": 1.5154,
      "step": 1216
    },
    {
      "epoch": 0.08006974028323766,
      "grad_norm": 0.6469805240631104,
      "learning_rate": 7.271774743739545e-06,
      "loss": 1.7022,
      "step": 1217
    },
    {
      "epoch": 0.08013553300327318,
      "grad_norm": 0.8975655436515808,
      "learning_rate": 7.216263653643435e-06,
      "loss": 1.8082,
      "step": 1218
    },
    {
      "epoch": 0.08020132572330872,
      "grad_norm": 0.625879168510437,
      "learning_rate": 7.16094877215725e-06,
      "loss": 1.472,
      "step": 1219
    },
    {
      "epoch": 0.08026711844334425,
      "grad_norm": 0.6286988258361816,
      "learning_rate": 7.105830352958142e-06,
      "loss": 1.5377,
      "step": 1220
    },
    {
      "epoch": 0.08033291116337977,
      "grad_norm": 0.6680283546447754,
      "learning_rate": 7.050908648822291e-06,
      "loss": 1.6344,
      "step": 1221
    },
    {
      "epoch": 0.0803987038834153,
      "grad_norm": 0.650658130645752,
      "learning_rate": 6.996183911623688e-06,
      "loss": 1.6916,
      "step": 1222
    },
    {
      "epoch": 0.08046449660345083,
      "grad_norm": 0.6374256014823914,
      "learning_rate": 6.941656392333046e-06,
      "loss": 1.5798,
      "step": 1223
    },
    {
      "epoch": 0.08053028932348635,
      "grad_norm": 0.6473804116249084,
      "learning_rate": 6.887326341016636e-06,
      "loss": 1.433,
      "step": 1224
    },
    {
      "epoch": 0.08059608204352188,
      "grad_norm": 0.6380757689476013,
      "learning_rate": 6.833194006835081e-06,
      "loss": 1.5023,
      "step": 1225
    },
    {
      "epoch": 0.0806618747635574,
      "grad_norm": 0.6916465759277344,
      "learning_rate": 6.779259638042318e-06,
      "loss": 1.6726,
      "step": 1226
    },
    {
      "epoch": 0.08072766748359295,
      "grad_norm": 0.6720482707023621,
      "learning_rate": 6.725523481984375e-06,
      "loss": 1.7212,
      "step": 1227
    },
    {
      "epoch": 0.08079346020362847,
      "grad_norm": 0.6906450986862183,
      "learning_rate": 6.671985785098278e-06,
      "loss": 1.4751,
      "step": 1228
    },
    {
      "epoch": 0.080859252923664,
      "grad_norm": 0.685138463973999,
      "learning_rate": 6.618646792910893e-06,
      "loss": 1.5658,
      "step": 1229
    },
    {
      "epoch": 0.08092504564369953,
      "grad_norm": 0.7090581059455872,
      "learning_rate": 6.565506750037836e-06,
      "loss": 1.5718,
      "step": 1230
    },
    {
      "epoch": 0.08099083836373505,
      "grad_norm": 0.6716530919075012,
      "learning_rate": 6.512565900182305e-06,
      "loss": 1.4331,
      "step": 1231
    },
    {
      "epoch": 0.08105663108377058,
      "grad_norm": 0.6960655450820923,
      "learning_rate": 6.459824486134014e-06,
      "loss": 1.6364,
      "step": 1232
    },
    {
      "epoch": 0.0811224238038061,
      "grad_norm": 0.7236138582229614,
      "learning_rate": 6.407282749768029e-06,
      "loss": 1.5542,
      "step": 1233
    },
    {
      "epoch": 0.08118821652384163,
      "grad_norm": 0.6596976518630981,
      "learning_rate": 6.354940932043713e-06,
      "loss": 1.5349,
      "step": 1234
    },
    {
      "epoch": 0.08125400924387717,
      "grad_norm": 0.6876804232597351,
      "learning_rate": 6.302799273003546e-06,
      "loss": 1.442,
      "step": 1235
    },
    {
      "epoch": 0.0813198019639127,
      "grad_norm": 0.710265576839447,
      "learning_rate": 6.2508580117720985e-06,
      "loss": 1.515,
      "step": 1236
    },
    {
      "epoch": 0.08138559468394822,
      "grad_norm": 0.7719532251358032,
      "learning_rate": 6.199117386554926e-06,
      "loss": 1.7442,
      "step": 1237
    },
    {
      "epoch": 0.08145138740398375,
      "grad_norm": 0.7328771948814392,
      "learning_rate": 6.147577634637414e-06,
      "loss": 1.569,
      "step": 1238
    },
    {
      "epoch": 0.08151718012401928,
      "grad_norm": 0.7373148202896118,
      "learning_rate": 6.096238992383752e-06,
      "loss": 1.5404,
      "step": 1239
    },
    {
      "epoch": 0.0815829728440548,
      "grad_norm": 0.8095849752426147,
      "learning_rate": 6.045101695235844e-06,
      "loss": 1.5851,
      "step": 1240
    },
    {
      "epoch": 0.08164876556409033,
      "grad_norm": 0.7829951047897339,
      "learning_rate": 5.994165977712174e-06,
      "loss": 1.5866,
      "step": 1241
    },
    {
      "epoch": 0.08171455828412587,
      "grad_norm": 0.7352398633956909,
      "learning_rate": 5.943432073406796e-06,
      "loss": 1.5109,
      "step": 1242
    },
    {
      "epoch": 0.0817803510041614,
      "grad_norm": 0.7571715116500854,
      "learning_rate": 5.892900214988245e-06,
      "loss": 1.5424,
      "step": 1243
    },
    {
      "epoch": 0.08184614372419692,
      "grad_norm": 0.806236207485199,
      "learning_rate": 5.842570634198452e-06,
      "loss": 1.6222,
      "step": 1244
    },
    {
      "epoch": 0.08191193644423245,
      "grad_norm": 0.894612729549408,
      "learning_rate": 5.792443561851685e-06,
      "loss": 1.4945,
      "step": 1245
    },
    {
      "epoch": 0.08197772916426797,
      "grad_norm": 0.8263707756996155,
      "learning_rate": 5.742519227833509e-06,
      "loss": 1.5567,
      "step": 1246
    },
    {
      "epoch": 0.0820435218843035,
      "grad_norm": 0.8109045624732971,
      "learning_rate": 5.692797861099719e-06,
      "loss": 1.4979,
      "step": 1247
    },
    {
      "epoch": 0.08210931460433903,
      "grad_norm": 1.0256260633468628,
      "learning_rate": 5.643279689675279e-06,
      "loss": 1.528,
      "step": 1248
    },
    {
      "epoch": 0.08217510732437455,
      "grad_norm": 0.9840426445007324,
      "learning_rate": 5.593964940653296e-06,
      "loss": 1.6114,
      "step": 1249
    },
    {
      "epoch": 0.08224090004441009,
      "grad_norm": 1.197351098060608,
      "learning_rate": 5.544853840193981e-06,
      "loss": 1.2792,
      "step": 1250
    },
    {
      "epoch": 0.08230669276444562,
      "grad_norm": 0.5352635383605957,
      "learning_rate": 5.495946613523567e-06,
      "loss": 1.7328,
      "step": 1251
    },
    {
      "epoch": 0.08237248548448114,
      "grad_norm": 0.5567160248756409,
      "learning_rate": 5.4472434849333396e-06,
      "loss": 1.6491,
      "step": 1252
    },
    {
      "epoch": 0.08243827820451667,
      "grad_norm": 0.6188229918479919,
      "learning_rate": 5.398744677778594e-06,
      "loss": 1.6258,
      "step": 1253
    },
    {
      "epoch": 0.0825040709245522,
      "grad_norm": 0.565366804599762,
      "learning_rate": 5.3504504144775535e-06,
      "loss": 1.7164,
      "step": 1254
    },
    {
      "epoch": 0.08256986364458772,
      "grad_norm": 0.5582414269447327,
      "learning_rate": 5.302360916510424e-06,
      "loss": 1.6116,
      "step": 1255
    },
    {
      "epoch": 0.08263565636462325,
      "grad_norm": 0.5899804830551147,
      "learning_rate": 5.25447640441834e-06,
      "loss": 1.6297,
      "step": 1256
    },
    {
      "epoch": 0.08270144908465878,
      "grad_norm": 0.6136707067489624,
      "learning_rate": 5.206797097802341e-06,
      "loss": 1.8456,
      "step": 1257
    },
    {
      "epoch": 0.08276724180469432,
      "grad_norm": 0.5709074139595032,
      "learning_rate": 5.1593232153223984e-06,
      "loss": 1.5469,
      "step": 1258
    },
    {
      "epoch": 0.08283303452472984,
      "grad_norm": 0.5794378519058228,
      "learning_rate": 5.112054974696395e-06,
      "loss": 1.594,
      "step": 1259
    },
    {
      "epoch": 0.08289882724476537,
      "grad_norm": 0.6078215837478638,
      "learning_rate": 5.064992592699136e-06,
      "loss": 1.7557,
      "step": 1260
    },
    {
      "epoch": 0.0829646199648009,
      "grad_norm": 0.6244221925735474,
      "learning_rate": 5.018136285161329e-06,
      "loss": 1.6409,
      "step": 1261
    },
    {
      "epoch": 0.08303041268483642,
      "grad_norm": 0.6333777904510498,
      "learning_rate": 4.9714862669686335e-06,
      "loss": 1.6453,
      "step": 1262
    },
    {
      "epoch": 0.08309620540487195,
      "grad_norm": 0.6196883320808411,
      "learning_rate": 4.925042752060638e-06,
      "loss": 1.539,
      "step": 1263
    },
    {
      "epoch": 0.08316199812490747,
      "grad_norm": 0.6235030293464661,
      "learning_rate": 4.87880595342991e-06,
      "loss": 1.6992,
      "step": 1264
    },
    {
      "epoch": 0.083227790844943,
      "grad_norm": 0.6125441789627075,
      "learning_rate": 4.832776083120982e-06,
      "loss": 1.5375,
      "step": 1265
    },
    {
      "epoch": 0.08329358356497854,
      "grad_norm": 0.6090467572212219,
      "learning_rate": 4.7869533522294395e-06,
      "loss": 1.5203,
      "step": 1266
    },
    {
      "epoch": 0.08335937628501407,
      "grad_norm": 0.6368080377578735,
      "learning_rate": 4.741337970900866e-06,
      "loss": 1.6075,
      "step": 1267
    },
    {
      "epoch": 0.0834251690050496,
      "grad_norm": 0.6555090546607971,
      "learning_rate": 4.695930148329958e-06,
      "loss": 1.666,
      "step": 1268
    },
    {
      "epoch": 0.08349096172508512,
      "grad_norm": 0.6124194264411926,
      "learning_rate": 4.650730092759542e-06,
      "loss": 1.5002,
      "step": 1269
    },
    {
      "epoch": 0.08355675444512065,
      "grad_norm": 0.6861993670463562,
      "learning_rate": 4.605738011479604e-06,
      "loss": 1.8197,
      "step": 1270
    },
    {
      "epoch": 0.08362254716515617,
      "grad_norm": 0.6512139439582825,
      "learning_rate": 4.560954110826337e-06,
      "loss": 1.5302,
      "step": 1271
    },
    {
      "epoch": 0.0836883398851917,
      "grad_norm": 0.6771849393844604,
      "learning_rate": 4.516378596181237e-06,
      "loss": 1.5859,
      "step": 1272
    },
    {
      "epoch": 0.08375413260522724,
      "grad_norm": 0.6660013198852539,
      "learning_rate": 4.472011671970083e-06,
      "loss": 1.6946,
      "step": 1273
    },
    {
      "epoch": 0.08381992532526276,
      "grad_norm": 0.6350880861282349,
      "learning_rate": 4.427853541662091e-06,
      "loss": 1.5967,
      "step": 1274
    },
    {
      "epoch": 0.08388571804529829,
      "grad_norm": 0.6567019820213318,
      "learning_rate": 4.383904407768907e-06,
      "loss": 1.5232,
      "step": 1275
    },
    {
      "epoch": 0.08395151076533382,
      "grad_norm": 0.6577147245407104,
      "learning_rate": 4.340164471843722e-06,
      "loss": 1.5343,
      "step": 1276
    },
    {
      "epoch": 0.08401730348536934,
      "grad_norm": 0.7278039455413818,
      "learning_rate": 4.296633934480337e-06,
      "loss": 1.6968,
      "step": 1277
    },
    {
      "epoch": 0.08408309620540487,
      "grad_norm": 0.7200085520744324,
      "learning_rate": 4.253312995312231e-06,
      "loss": 1.5763,
      "step": 1278
    },
    {
      "epoch": 0.0841488889254404,
      "grad_norm": 0.6737844944000244,
      "learning_rate": 4.210201853011652e-06,
      "loss": 1.5614,
      "step": 1279
    },
    {
      "epoch": 0.08421468164547592,
      "grad_norm": 0.7159338593482971,
      "learning_rate": 4.167300705288718e-06,
      "loss": 1.7838,
      "step": 1280
    },
    {
      "epoch": 0.08428047436551146,
      "grad_norm": 0.7275623679161072,
      "learning_rate": 4.12460974889049e-06,
      "loss": 1.7268,
      "step": 1281
    },
    {
      "epoch": 0.08434626708554699,
      "grad_norm": 0.6852192878723145,
      "learning_rate": 4.082129179600097e-06,
      "loss": 1.4123,
      "step": 1282
    },
    {
      "epoch": 0.08441205980558251,
      "grad_norm": 0.7260058522224426,
      "learning_rate": 4.039859192235779e-06,
      "loss": 1.5387,
      "step": 1283
    },
    {
      "epoch": 0.08447785252561804,
      "grad_norm": 0.7276716828346252,
      "learning_rate": 3.99779998065008e-06,
      "loss": 1.6186,
      "step": 1284
    },
    {
      "epoch": 0.08454364524565357,
      "grad_norm": 0.8000338077545166,
      "learning_rate": 3.955951737728902e-06,
      "loss": 1.544,
      "step": 1285
    },
    {
      "epoch": 0.0846094379656891,
      "grad_norm": 0.7734615206718445,
      "learning_rate": 3.914314655390633e-06,
      "loss": 1.628,
      "step": 1286
    },
    {
      "epoch": 0.08467523068572462,
      "grad_norm": 0.742764949798584,
      "learning_rate": 3.872888924585255e-06,
      "loss": 1.5357,
      "step": 1287
    },
    {
      "epoch": 0.08474102340576015,
      "grad_norm": 0.7545043230056763,
      "learning_rate": 3.831674735293506e-06,
      "loss": 1.6103,
      "step": 1288
    },
    {
      "epoch": 0.08480681612579569,
      "grad_norm": 0.7599538564682007,
      "learning_rate": 3.790672276525936e-06,
      "loss": 1.5277,
      "step": 1289
    },
    {
      "epoch": 0.08487260884583121,
      "grad_norm": 0.7628370523452759,
      "learning_rate": 3.749881736322136e-06,
      "loss": 1.5536,
      "step": 1290
    },
    {
      "epoch": 0.08493840156586674,
      "grad_norm": 0.7717123031616211,
      "learning_rate": 3.7093033017497946e-06,
      "loss": 1.6771,
      "step": 1291
    },
    {
      "epoch": 0.08500419428590227,
      "grad_norm": 0.7848740220069885,
      "learning_rate": 3.668937158903901e-06,
      "loss": 1.4859,
      "step": 1292
    },
    {
      "epoch": 0.08506998700593779,
      "grad_norm": 0.8771865367889404,
      "learning_rate": 3.6287834929058295e-06,
      "loss": 1.5999,
      "step": 1293
    },
    {
      "epoch": 0.08513577972597332,
      "grad_norm": 0.8170807361602783,
      "learning_rate": 3.5888424879025495e-06,
      "loss": 1.4485,
      "step": 1294
    },
    {
      "epoch": 0.08520157244600884,
      "grad_norm": 0.8586212396621704,
      "learning_rate": 3.5491143270657446e-06,
      "loss": 1.6085,
      "step": 1295
    },
    {
      "epoch": 0.08526736516604437,
      "grad_norm": 0.8808199763298035,
      "learning_rate": 3.5095991925909845e-06,
      "loss": 1.4775,
      "step": 1296
    },
    {
      "epoch": 0.08533315788607991,
      "grad_norm": 0.8514795303344727,
      "learning_rate": 3.470297265696887e-06,
      "loss": 1.4806,
      "step": 1297
    },
    {
      "epoch": 0.08539895060611544,
      "grad_norm": 0.9353112578392029,
      "learning_rate": 3.4312087266242963e-06,
      "loss": 1.4445,
      "step": 1298
    },
    {
      "epoch": 0.08546474332615096,
      "grad_norm": 0.9689911007881165,
      "learning_rate": 3.3923337546354297e-06,
      "loss": 1.2047,
      "step": 1299
    },
    {
      "epoch": 0.08553053604618649,
      "grad_norm": 1.2559093236923218,
      "learning_rate": 3.3536725280130744e-06,
      "loss": 1.2073,
      "step": 1300
    },
    {
      "epoch": 0.08559632876622202,
      "grad_norm": 0.5083854794502258,
      "learning_rate": 3.315225224059809e-06,
      "loss": 1.5963,
      "step": 1301
    },
    {
      "epoch": 0.08566212148625754,
      "grad_norm": 0.5269922018051147,
      "learning_rate": 3.2769920190971027e-06,
      "loss": 1.5835,
      "step": 1302
    },
    {
      "epoch": 0.08572791420629307,
      "grad_norm": 0.5758770704269409,
      "learning_rate": 3.2389730884645807e-06,
      "loss": 1.6247,
      "step": 1303
    },
    {
      "epoch": 0.08579370692632861,
      "grad_norm": 0.5545147061347961,
      "learning_rate": 3.2011686065191895e-06,
      "loss": 1.6925,
      "step": 1304
    },
    {
      "epoch": 0.08585949964636413,
      "grad_norm": 0.5974170565605164,
      "learning_rate": 3.163578746634388e-06,
      "loss": 1.7166,
      "step": 1305
    },
    {
      "epoch": 0.08592529236639966,
      "grad_norm": 0.5378652215003967,
      "learning_rate": 3.1262036811993856e-06,
      "loss": 1.5484,
      "step": 1306
    },
    {
      "epoch": 0.08599108508643519,
      "grad_norm": 0.6244592666625977,
      "learning_rate": 3.0890435816183226e-06,
      "loss": 1.8135,
      "step": 1307
    },
    {
      "epoch": 0.08605687780647071,
      "grad_norm": 0.5631170272827148,
      "learning_rate": 3.0520986183095014e-06,
      "loss": 1.6121,
      "step": 1308
    },
    {
      "epoch": 0.08612267052650624,
      "grad_norm": 0.5755576491355896,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.5121,
      "step": 1309
    },
    {
      "epoch": 0.08618846324654177,
      "grad_norm": 0.6194965839385986,
      "learning_rate": 2.9788547772478416e-06,
      "loss": 1.5921,
      "step": 1310
    },
    {
      "epoch": 0.08625425596657729,
      "grad_norm": 0.6144405603408813,
      "learning_rate": 2.9425562353953604e-06,
      "loss": 1.683,
      "step": 1311
    },
    {
      "epoch": 0.08632004868661283,
      "grad_norm": 0.6393880844116211,
      "learning_rate": 2.9064735016142873e-06,
      "loss": 1.6299,
      "step": 1312
    },
    {
      "epoch": 0.08638584140664836,
      "grad_norm": 0.6104136109352112,
      "learning_rate": 2.870606741382059e-06,
      "loss": 1.6589,
      "step": 1313
    },
    {
      "epoch": 0.08645163412668389,
      "grad_norm": 0.6379032135009766,
      "learning_rate": 2.83495611918565e-06,
      "loss": 1.5917,
      "step": 1314
    },
    {
      "epoch": 0.08651742684671941,
      "grad_norm": 0.6215010285377502,
      "learning_rate": 2.7995217985208098e-06,
      "loss": 1.5812,
      "step": 1315
    },
    {
      "epoch": 0.08658321956675494,
      "grad_norm": 0.6086938977241516,
      "learning_rate": 2.7643039418913e-06,
      "loss": 1.5549,
      "step": 1316
    },
    {
      "epoch": 0.08664901228679046,
      "grad_norm": 0.6653648018836975,
      "learning_rate": 2.729302710808196e-06,
      "loss": 1.4754,
      "step": 1317
    },
    {
      "epoch": 0.08671480500682599,
      "grad_norm": 0.6537653803825378,
      "learning_rate": 2.6945182657891034e-06,
      "loss": 1.7067,
      "step": 1318
    },
    {
      "epoch": 0.08678059772686152,
      "grad_norm": 0.6637395620346069,
      "learning_rate": 2.6599507663574384e-06,
      "loss": 1.5416,
      "step": 1319
    },
    {
      "epoch": 0.08684639044689706,
      "grad_norm": 0.6458157300949097,
      "learning_rate": 2.6256003710416864e-06,
      "loss": 1.5356,
      "step": 1320
    },
    {
      "epoch": 0.08691218316693258,
      "grad_norm": 0.6820713877677917,
      "learning_rate": 2.5914672373746674e-06,
      "loss": 1.797,
      "step": 1321
    },
    {
      "epoch": 0.08697797588696811,
      "grad_norm": 0.6089844107627869,
      "learning_rate": 2.5575515218928592e-06,
      "loss": 1.5584,
      "step": 1322
    },
    {
      "epoch": 0.08704376860700364,
      "grad_norm": 0.7012182474136353,
      "learning_rate": 2.5238533801356324e-06,
      "loss": 1.6888,
      "step": 1323
    },
    {
      "epoch": 0.08710956132703916,
      "grad_norm": 0.6670122742652893,
      "learning_rate": 2.490372966644544e-06,
      "loss": 1.6455,
      "step": 1324
    },
    {
      "epoch": 0.08717535404707469,
      "grad_norm": 0.6817238330841064,
      "learning_rate": 2.457110434962645e-06,
      "loss": 1.575,
      "step": 1325
    },
    {
      "epoch": 0.08724114676711021,
      "grad_norm": 0.6881915926933289,
      "learning_rate": 2.424065937633768e-06,
      "loss": 1.7438,
      "step": 1326
    },
    {
      "epoch": 0.08730693948714574,
      "grad_norm": 0.6820628643035889,
      "learning_rate": 2.3912396262018357e-06,
      "loss": 1.5352,
      "step": 1327
    },
    {
      "epoch": 0.08737273220718128,
      "grad_norm": 0.7019688487052917,
      "learning_rate": 2.3586316512101416e-06,
      "loss": 1.6225,
      "step": 1328
    },
    {
      "epoch": 0.0874385249272168,
      "grad_norm": 0.6856455206871033,
      "learning_rate": 2.3262421622006868e-06,
      "loss": 1.7121,
      "step": 1329
    },
    {
      "epoch": 0.08750431764725233,
      "grad_norm": 0.7237346768379211,
      "learning_rate": 2.29407130771348e-06,
      "loss": 1.6317,
      "step": 1330
    },
    {
      "epoch": 0.08757011036728786,
      "grad_norm": 0.728233814239502,
      "learning_rate": 2.26211923528587e-06,
      "loss": 1.5668,
      "step": 1331
    },
    {
      "epoch": 0.08763590308732339,
      "grad_norm": 0.6574517488479614,
      "learning_rate": 2.2303860914518306e-06,
      "loss": 1.496,
      "step": 1332
    },
    {
      "epoch": 0.08770169580735891,
      "grad_norm": 0.7108151912689209,
      "learning_rate": 2.1988720217413494e-06,
      "loss": 1.4753,
      "step": 1333
    },
    {
      "epoch": 0.08776748852739444,
      "grad_norm": 0.7035221457481384,
      "learning_rate": 2.1675771706797132e-06,
      "loss": 1.4783,
      "step": 1334
    },
    {
      "epoch": 0.08783328124742998,
      "grad_norm": 0.713604211807251,
      "learning_rate": 2.136501681786862e-06,
      "loss": 1.6723,
      "step": 1335
    },
    {
      "epoch": 0.0878990739674655,
      "grad_norm": 0.7584853768348694,
      "learning_rate": 2.10564569757673e-06,
      "loss": 1.457,
      "step": 1336
    },
    {
      "epoch": 0.08796486668750103,
      "grad_norm": 0.7580429911613464,
      "learning_rate": 2.0750093595565733e-06,
      "loss": 1.597,
      "step": 1337
    },
    {
      "epoch": 0.08803065940753656,
      "grad_norm": 0.7131192684173584,
      "learning_rate": 2.0445928082263645e-06,
      "loss": 1.5909,
      "step": 1338
    },
    {
      "epoch": 0.08809645212757208,
      "grad_norm": 0.7904130816459656,
      "learning_rate": 2.01439618307811e-06,
      "loss": 1.5464,
      "step": 1339
    },
    {
      "epoch": 0.08816224484760761,
      "grad_norm": 0.8404092192649841,
      "learning_rate": 1.984419622595224e-06,
      "loss": 1.616,
      "step": 1340
    },
    {
      "epoch": 0.08822803756764314,
      "grad_norm": 0.7262335419654846,
      "learning_rate": 1.954663264251888e-06,
      "loss": 1.5096,
      "step": 1341
    },
    {
      "epoch": 0.08829383028767866,
      "grad_norm": 0.8143447041511536,
      "learning_rate": 1.925127244512426e-06,
      "loss": 1.4777,
      "step": 1342
    },
    {
      "epoch": 0.0883596230077142,
      "grad_norm": 0.8100689053535461,
      "learning_rate": 1.895811698830685e-06,
      "loss": 1.4577,
      "step": 1343
    },
    {
      "epoch": 0.08842541572774973,
      "grad_norm": 0.8460294008255005,
      "learning_rate": 1.8667167616493896e-06,
      "loss": 1.5934,
      "step": 1344
    },
    {
      "epoch": 0.08849120844778526,
      "grad_norm": 0.8502865433692932,
      "learning_rate": 1.8378425663995559e-06,
      "loss": 1.435,
      "step": 1345
    },
    {
      "epoch": 0.08855700116782078,
      "grad_norm": 0.9002560377120972,
      "learning_rate": 1.8091892454998594e-06,
      "loss": 1.459,
      "step": 1346
    },
    {
      "epoch": 0.08862279388785631,
      "grad_norm": 0.8573265075683594,
      "learning_rate": 1.7807569303560367e-06,
      "loss": 1.5029,
      "step": 1347
    },
    {
      "epoch": 0.08868858660789183,
      "grad_norm": 0.8953090310096741,
      "learning_rate": 1.7525457513602683e-06,
      "loss": 1.4322,
      "step": 1348
    },
    {
      "epoch": 0.08875437932792736,
      "grad_norm": 0.9949759244918823,
      "learning_rate": 1.7245558378906013e-06,
      "loss": 1.4967,
      "step": 1349
    },
    {
      "epoch": 0.08882017204796289,
      "grad_norm": 1.1277623176574707,
      "learning_rate": 1.6967873183103556e-06,
      "loss": 1.0865,
      "step": 1350
    },
    {
      "epoch": 0.08888596476799843,
      "grad_norm": 0.5030365586280823,
      "learning_rate": 1.6692403199675078e-06,
      "loss": 1.6924,
      "step": 1351
    },
    {
      "epoch": 0.08895175748803395,
      "grad_norm": 0.5737792253494263,
      "learning_rate": 1.641914969194147e-06,
      "loss": 1.7576,
      "step": 1352
    },
    {
      "epoch": 0.08901755020806948,
      "grad_norm": 0.5396728515625,
      "learning_rate": 1.6148113913058427e-06,
      "loss": 1.4578,
      "step": 1353
    },
    {
      "epoch": 0.089083342928105,
      "grad_norm": 0.535679280757904,
      "learning_rate": 1.587929710601127e-06,
      "loss": 1.5822,
      "step": 1354
    },
    {
      "epoch": 0.08914913564814053,
      "grad_norm": 0.5770007371902466,
      "learning_rate": 1.5612700503608968e-06,
      "loss": 1.7699,
      "step": 1355
    },
    {
      "epoch": 0.08921492836817606,
      "grad_norm": 0.5972631573677063,
      "learning_rate": 1.5348325328478408e-06,
      "loss": 1.6394,
      "step": 1356
    },
    {
      "epoch": 0.08928072108821158,
      "grad_norm": 0.5818490386009216,
      "learning_rate": 1.5086172793059017e-06,
      "loss": 1.7062,
      "step": 1357
    },
    {
      "epoch": 0.08934651380824711,
      "grad_norm": 0.6017389893531799,
      "learning_rate": 1.4826244099596986e-06,
      "loss": 1.6664,
      "step": 1358
    },
    {
      "epoch": 0.08941230652828265,
      "grad_norm": 0.6041622161865234,
      "learning_rate": 1.4568540440139777e-06,
      "loss": 1.7427,
      "step": 1359
    },
    {
      "epoch": 0.08947809924831818,
      "grad_norm": 0.8328604698181152,
      "learning_rate": 1.4313062996530847e-06,
      "loss": 1.5239,
      "step": 1360
    },
    {
      "epoch": 0.0895438919683537,
      "grad_norm": 0.6048279404640198,
      "learning_rate": 1.4059812940404093e-06,
      "loss": 1.7337,
      "step": 1361
    },
    {
      "epoch": 0.08960968468838923,
      "grad_norm": 0.6227853894233704,
      "learning_rate": 1.3808791433178369e-06,
      "loss": 1.451,
      "step": 1362
    },
    {
      "epoch": 0.08967547740842476,
      "grad_norm": 0.5862622857093811,
      "learning_rate": 1.3559999626052477e-06,
      "loss": 1.4699,
      "step": 1363
    },
    {
      "epoch": 0.08974127012846028,
      "grad_norm": 0.6687149405479431,
      "learning_rate": 1.33134386599994e-06,
      "loss": 1.5708,
      "step": 1364
    },
    {
      "epoch": 0.08980706284849581,
      "grad_norm": 0.65779709815979,
      "learning_rate": 1.3069109665761693e-06,
      "loss": 1.6006,
      "step": 1365
    },
    {
      "epoch": 0.08987285556853135,
      "grad_norm": 0.6536720991134644,
      "learning_rate": 1.2827013763845707e-06,
      "loss": 1.6428,
      "step": 1366
    },
    {
      "epoch": 0.08993864828856687,
      "grad_norm": 0.6218163967132568,
      "learning_rate": 1.2587152064516827e-06,
      "loss": 1.5861,
      "step": 1367
    },
    {
      "epoch": 0.0900044410086024,
      "grad_norm": 0.6612845063209534,
      "learning_rate": 1.2349525667794293e-06,
      "loss": 1.7162,
      "step": 1368
    },
    {
      "epoch": 0.09007023372863793,
      "grad_norm": 0.6771432757377625,
      "learning_rate": 1.211413566344599e-06,
      "loss": 1.7318,
      "step": 1369
    },
    {
      "epoch": 0.09013602644867345,
      "grad_norm": 0.6161140203475952,
      "learning_rate": 1.1880983130983626e-06,
      "loss": 1.5076,
      "step": 1370
    },
    {
      "epoch": 0.09020181916870898,
      "grad_norm": 0.6894911527633667,
      "learning_rate": 1.1650069139657826e-06,
      "loss": 1.7694,
      "step": 1371
    },
    {
      "epoch": 0.0902676118887445,
      "grad_norm": 0.6760029196739197,
      "learning_rate": 1.1421394748453108e-06,
      "loss": 1.6852,
      "step": 1372
    },
    {
      "epoch": 0.09033340460878003,
      "grad_norm": 0.716984748840332,
      "learning_rate": 1.1194961006082972e-06,
      "loss": 1.4935,
      "step": 1373
    },
    {
      "epoch": 0.09039919732881557,
      "grad_norm": 0.6679751873016357,
      "learning_rate": 1.0970768950985199e-06,
      "loss": 1.7724,
      "step": 1374
    },
    {
      "epoch": 0.0904649900488511,
      "grad_norm": 0.6891523599624634,
      "learning_rate": 1.074881961131724e-06,
      "loss": 1.5886,
      "step": 1375
    },
    {
      "epoch": 0.09053078276888663,
      "grad_norm": 0.6760563254356384,
      "learning_rate": 1.0529114004951047e-06,
      "loss": 1.632,
      "step": 1376
    },
    {
      "epoch": 0.09059657548892215,
      "grad_norm": 0.6902234554290771,
      "learning_rate": 1.0311653139468969e-06,
      "loss": 1.6434,
      "step": 1377
    },
    {
      "epoch": 0.09066236820895768,
      "grad_norm": 0.6432000398635864,
      "learning_rate": 1.0096438012158539e-06,
      "loss": 1.5294,
      "step": 1378
    },
    {
      "epoch": 0.0907281609289932,
      "grad_norm": 0.6482113599777222,
      "learning_rate": 9.883469610008577e-07,
      "loss": 1.5759,
      "step": 1379
    },
    {
      "epoch": 0.09079395364902873,
      "grad_norm": 0.7423402070999146,
      "learning_rate": 9.672748909703934e-07,
      "loss": 1.697,
      "step": 1380
    },
    {
      "epoch": 0.09085974636906426,
      "grad_norm": 0.7108086943626404,
      "learning_rate": 9.46427687762158e-07,
      "loss": 1.6335,
      "step": 1381
    },
    {
      "epoch": 0.0909255390890998,
      "grad_norm": 0.6593964695930481,
      "learning_rate": 9.258054469825972e-07,
      "loss": 1.5451,
      "step": 1382
    },
    {
      "epoch": 0.09099133180913532,
      "grad_norm": 0.7182891964912415,
      "learning_rate": 9.054082632064642e-07,
      "loss": 1.4827,
      "step": 1383
    },
    {
      "epoch": 0.09105712452917085,
      "grad_norm": 0.7174695730209351,
      "learning_rate": 8.852362299763772e-07,
      "loss": 1.652,
      "step": 1384
    },
    {
      "epoch": 0.09112291724920638,
      "grad_norm": 0.7062637209892273,
      "learning_rate": 8.652894398024136e-07,
      "loss": 1.478,
      "step": 1385
    },
    {
      "epoch": 0.0911887099692419,
      "grad_norm": 0.737032949924469,
      "learning_rate": 8.455679841616659e-07,
      "loss": 1.6173,
      "step": 1386
    },
    {
      "epoch": 0.09125450268927743,
      "grad_norm": 0.8026255369186401,
      "learning_rate": 8.260719534978368e-07,
      "loss": 1.5671,
      "step": 1387
    },
    {
      "epoch": 0.09132029540931295,
      "grad_norm": 0.7752266526222229,
      "learning_rate": 8.06801437220811e-07,
      "loss": 1.7083,
      "step": 1388
    },
    {
      "epoch": 0.0913860881293485,
      "grad_norm": 0.7705795764923096,
      "learning_rate": 7.877565237062623e-07,
      "loss": 1.4577,
      "step": 1389
    },
    {
      "epoch": 0.09145188084938402,
      "grad_norm": 0.799823522567749,
      "learning_rate": 7.689373002952305e-07,
      "loss": 1.7553,
      "step": 1390
    },
    {
      "epoch": 0.09151767356941955,
      "grad_norm": 0.7618536353111267,
      "learning_rate": 7.503438532937168e-07,
      "loss": 1.4927,
      "step": 1391
    },
    {
      "epoch": 0.09158346628945507,
      "grad_norm": 0.9011740684509277,
      "learning_rate": 7.319762679723174e-07,
      "loss": 1.6078,
      "step": 1392
    },
    {
      "epoch": 0.0916492590094906,
      "grad_norm": 0.8119943737983704,
      "learning_rate": 7.138346285658071e-07,
      "loss": 1.3945,
      "step": 1393
    },
    {
      "epoch": 0.09171505172952613,
      "grad_norm": 0.8282358050346375,
      "learning_rate": 6.959190182727615e-07,
      "loss": 1.6105,
      "step": 1394
    },
    {
      "epoch": 0.09178084444956165,
      "grad_norm": 0.8864809274673462,
      "learning_rate": 6.782295192551691e-07,
      "loss": 1.6667,
      "step": 1395
    },
    {
      "epoch": 0.09184663716959718,
      "grad_norm": 0.8417677879333496,
      "learning_rate": 6.607662126380587e-07,
      "loss": 1.513,
      "step": 1396
    },
    {
      "epoch": 0.09191242988963272,
      "grad_norm": 0.8300901651382446,
      "learning_rate": 6.43529178509139e-07,
      "loss": 1.3415,
      "step": 1397
    },
    {
      "epoch": 0.09197822260966824,
      "grad_norm": 0.9474268555641174,
      "learning_rate": 6.265184959184101e-07,
      "loss": 1.56,
      "step": 1398
    },
    {
      "epoch": 0.09204401532970377,
      "grad_norm": 1.1085734367370605,
      "learning_rate": 6.097342428778185e-07,
      "loss": 1.4722,
      "step": 1399
    },
    {
      "epoch": 0.0921098080497393,
      "grad_norm": 1.2595409154891968,
      "learning_rate": 5.931764963608866e-07,
      "loss": 1.2695,
      "step": 1400
    },
    {
      "epoch": 0.09217560076977482,
      "grad_norm": 0.4739496409893036,
      "learning_rate": 5.768453323023615e-07,
      "loss": 1.5488,
      "step": 1401
    },
    {
      "epoch": 0.09224139348981035,
      "grad_norm": 0.550033688545227,
      "learning_rate": 5.60740825597883e-07,
      "loss": 1.761,
      "step": 1402
    },
    {
      "epoch": 0.09230718620984588,
      "grad_norm": 0.5993760824203491,
      "learning_rate": 5.448630501036112e-07,
      "loss": 1.8681,
      "step": 1403
    },
    {
      "epoch": 0.0923729789298814,
      "grad_norm": 0.5492662787437439,
      "learning_rate": 5.292120786359267e-07,
      "loss": 1.6244,
      "step": 1404
    },
    {
      "epoch": 0.09243877164991694,
      "grad_norm": 0.5324863195419312,
      "learning_rate": 5.137879829710424e-07,
      "loss": 1.579,
      "step": 1405
    },
    {
      "epoch": 0.09250456436995247,
      "grad_norm": 0.5814157724380493,
      "learning_rate": 4.985908338447476e-07,
      "loss": 1.6163,
      "step": 1406
    },
    {
      "epoch": 0.092570357089988,
      "grad_norm": 0.5995772480964661,
      "learning_rate": 4.836207009519977e-07,
      "loss": 1.657,
      "step": 1407
    },
    {
      "epoch": 0.09263614981002352,
      "grad_norm": 0.6003164649009705,
      "learning_rate": 4.688776529466754e-07,
      "loss": 1.6132,
      "step": 1408
    },
    {
      "epoch": 0.09270194253005905,
      "grad_norm": 0.5804505348205566,
      "learning_rate": 4.543617574412184e-07,
      "loss": 1.5816,
      "step": 1409
    },
    {
      "epoch": 0.09276773525009457,
      "grad_norm": 0.5852599143981934,
      "learning_rate": 4.4007308100633136e-07,
      "loss": 1.6799,
      "step": 1410
    },
    {
      "epoch": 0.0928335279701301,
      "grad_norm": 0.6366518139839172,
      "learning_rate": 4.2601168917069114e-07,
      "loss": 1.7649,
      "step": 1411
    },
    {
      "epoch": 0.09289932069016563,
      "grad_norm": 0.5954082608222961,
      "learning_rate": 4.121776464206251e-07,
      "loss": 1.6245,
      "step": 1412
    },
    {
      "epoch": 0.09296511341020117,
      "grad_norm": 0.6634638905525208,
      "learning_rate": 3.9857101619982797e-07,
      "loss": 1.7316,
      "step": 1413
    },
    {
      "epoch": 0.09303090613023669,
      "grad_norm": 0.6314533352851868,
      "learning_rate": 3.851918609090677e-07,
      "loss": 1.6691,
      "step": 1414
    },
    {
      "epoch": 0.09309669885027222,
      "grad_norm": 0.6321081519126892,
      "learning_rate": 3.720402419058966e-07,
      "loss": 1.4713,
      "step": 1415
    },
    {
      "epoch": 0.09316249157030775,
      "grad_norm": 0.6127361059188843,
      "learning_rate": 3.5911621950438514e-07,
      "loss": 1.5676,
      "step": 1416
    },
    {
      "epoch": 0.09322828429034327,
      "grad_norm": 0.6360514760017395,
      "learning_rate": 3.464198529748108e-07,
      "loss": 1.5381,
      "step": 1417
    },
    {
      "epoch": 0.0932940770103788,
      "grad_norm": 0.6701735258102417,
      "learning_rate": 3.339512005434309e-07,
      "loss": 1.6244,
      "step": 1418
    },
    {
      "epoch": 0.09335986973041432,
      "grad_norm": 0.6812663674354553,
      "learning_rate": 3.2171031939217666e-07,
      "loss": 1.6387,
      "step": 1419
    },
    {
      "epoch": 0.09342566245044986,
      "grad_norm": 0.6711384057998657,
      "learning_rate": 3.0969726565842074e-07,
      "loss": 1.5678,
      "step": 1420
    },
    {
      "epoch": 0.09349145517048539,
      "grad_norm": 0.7010900974273682,
      "learning_rate": 2.979120944346936e-07,
      "loss": 1.6543,
      "step": 1421
    },
    {
      "epoch": 0.09355724789052092,
      "grad_norm": 0.6457873582839966,
      "learning_rate": 2.863548597684562e-07,
      "loss": 1.6793,
      "step": 1422
    },
    {
      "epoch": 0.09362304061055644,
      "grad_norm": 0.6923664808273315,
      "learning_rate": 2.750256146618335e-07,
      "loss": 1.5873,
      "step": 1423
    },
    {
      "epoch": 0.09368883333059197,
      "grad_norm": 0.6824166178703308,
      "learning_rate": 2.639244110713701e-07,
      "loss": 1.5315,
      "step": 1424
    },
    {
      "epoch": 0.0937546260506275,
      "grad_norm": 0.6563824415206909,
      "learning_rate": 2.5305129990781387e-07,
      "loss": 1.604,
      "step": 1425
    },
    {
      "epoch": 0.09382041877066302,
      "grad_norm": 0.6801635026931763,
      "learning_rate": 2.424063310358604e-07,
      "loss": 1.5743,
      "step": 1426
    },
    {
      "epoch": 0.09388621149069855,
      "grad_norm": 0.7070457935333252,
      "learning_rate": 2.319895532739369e-07,
      "loss": 1.5733,
      "step": 1427
    },
    {
      "epoch": 0.09395200421073409,
      "grad_norm": 0.661389172077179,
      "learning_rate": 2.218010143939575e-07,
      "loss": 1.4381,
      "step": 1428
    },
    {
      "epoch": 0.09401779693076961,
      "grad_norm": 0.7122326493263245,
      "learning_rate": 2.1184076112114038e-07,
      "loss": 1.5635,
      "step": 1429
    },
    {
      "epoch": 0.09408358965080514,
      "grad_norm": 0.6946954131126404,
      "learning_rate": 2.0210883913376334e-07,
      "loss": 1.558,
      "step": 1430
    },
    {
      "epoch": 0.09414938237084067,
      "grad_norm": 0.7263256311416626,
      "learning_rate": 1.9260529306296404e-07,
      "loss": 1.6198,
      "step": 1431
    },
    {
      "epoch": 0.0942151750908762,
      "grad_norm": 0.7192728519439697,
      "learning_rate": 1.833301664925402e-07,
      "loss": 1.6123,
      "step": 1432
    },
    {
      "epoch": 0.09428096781091172,
      "grad_norm": 0.6984828114509583,
      "learning_rate": 1.742835019587441e-07,
      "loss": 1.5176,
      "step": 1433
    },
    {
      "epoch": 0.09434676053094725,
      "grad_norm": 0.7076666951179504,
      "learning_rate": 1.6546534095007172e-07,
      "loss": 1.5039,
      "step": 1434
    },
    {
      "epoch": 0.09441255325098277,
      "grad_norm": 0.7117300629615784,
      "learning_rate": 1.5687572390711835e-07,
      "loss": 1.5169,
      "step": 1435
    },
    {
      "epoch": 0.09447834597101831,
      "grad_norm": 0.7957559823989868,
      "learning_rate": 1.4851469022234e-07,
      "loss": 1.6177,
      "step": 1436
    },
    {
      "epoch": 0.09454413869105384,
      "grad_norm": 0.7519058585166931,
      "learning_rate": 1.403822782399089e-07,
      "loss": 1.5329,
      "step": 1437
    },
    {
      "epoch": 0.09460993141108937,
      "grad_norm": 0.7622172236442566,
      "learning_rate": 1.324785252555194e-07,
      "loss": 1.3998,
      "step": 1438
    },
    {
      "epoch": 0.09467572413112489,
      "grad_norm": 0.767694354057312,
      "learning_rate": 1.2480346751622686e-07,
      "loss": 1.5882,
      "step": 1439
    },
    {
      "epoch": 0.09474151685116042,
      "grad_norm": 0.7719693183898926,
      "learning_rate": 1.1735714022027555e-07,
      "loss": 1.3252,
      "step": 1440
    },
    {
      "epoch": 0.09480730957119594,
      "grad_norm": 0.7426503896713257,
      "learning_rate": 1.1013957751693782e-07,
      "loss": 1.5377,
      "step": 1441
    },
    {
      "epoch": 0.09487310229123147,
      "grad_norm": 0.7355154156684875,
      "learning_rate": 1.0315081250636405e-07,
      "loss": 1.424,
      "step": 1442
    },
    {
      "epoch": 0.094938895011267,
      "grad_norm": 0.8036089539527893,
      "learning_rate": 9.63908772394162e-08,
      "loss": 1.605,
      "step": 1443
    },
    {
      "epoch": 0.09500468773130254,
      "grad_norm": 0.878667950630188,
      "learning_rate": 8.985980271754013e-08,
      "loss": 1.5414,
      "step": 1444
    },
    {
      "epoch": 0.09507048045133806,
      "grad_norm": 0.9153657555580139,
      "learning_rate": 8.355761889260461e-08,
      "loss": 1.5492,
      "step": 1445
    },
    {
      "epoch": 0.09513627317137359,
      "grad_norm": 0.8688125014305115,
      "learning_rate": 7.748435466678471e-08,
      "loss": 1.525,
      "step": 1446
    },
    {
      "epoch": 0.09520206589140912,
      "grad_norm": 0.9009868502616882,
      "learning_rate": 7.164003789240648e-08,
      "loss": 1.4909,
      "step": 1447
    },
    {
      "epoch": 0.09526785861144464,
      "grad_norm": 0.947225034236908,
      "learning_rate": 6.602469537183021e-08,
      "loss": 1.4103,
      "step": 1448
    },
    {
      "epoch": 0.09533365133148017,
      "grad_norm": 1.0838871002197266,
      "learning_rate": 6.063835285733955e-08,
      "loss": 1.5726,
      "step": 1449
    },
    {
      "epoch": 0.0953994440515157,
      "grad_norm": 1.4739456176757812,
      "learning_rate": 5.5481035050991556e-08,
      "loss": 1.0046,
      "step": 1450
    },
    {
      "epoch": 0.09546523677155123,
      "grad_norm": 0.5145021677017212,
      "learning_rate": 5.0552765604544584e-08,
      "loss": 1.7385,
      "step": 1451
    },
    {
      "epoch": 0.09553102949158676,
      "grad_norm": 0.5236654877662659,
      "learning_rate": 4.585356711931388e-08,
      "loss": 1.7165,
      "step": 1452
    },
    {
      "epoch": 0.09559682221162229,
      "grad_norm": 0.5186619758605957,
      "learning_rate": 4.138346114608283e-08,
      "loss": 1.4829,
      "step": 1453
    },
    {
      "epoch": 0.09566261493165781,
      "grad_norm": 0.5814453959465027,
      "learning_rate": 3.7142468185014104e-08,
      "loss": 1.6926,
      "step": 1454
    },
    {
      "epoch": 0.09572840765169334,
      "grad_norm": 0.5442173480987549,
      "learning_rate": 3.313060768553866e-08,
      "loss": 1.6347,
      "step": 1455
    },
    {
      "epoch": 0.09579420037172887,
      "grad_norm": 0.5665310025215149,
      "learning_rate": 2.9347898046266918e-08,
      "loss": 1.5986,
      "step": 1456
    },
    {
      "epoch": 0.09585999309176439,
      "grad_norm": 0.5988601446151733,
      "learning_rate": 2.5794356614922134e-08,
      "loss": 1.8659,
      "step": 1457
    },
    {
      "epoch": 0.09592578581179992,
      "grad_norm": 0.5948224067687988,
      "learning_rate": 2.2469999688246035e-08,
      "loss": 1.7225,
      "step": 1458
    },
    {
      "epoch": 0.09599157853183546,
      "grad_norm": 0.6055050492286682,
      "learning_rate": 1.937484251192112e-08,
      "loss": 1.6482,
      "step": 1459
    },
    {
      "epoch": 0.09605737125187098,
      "grad_norm": 0.6156303286552429,
      "learning_rate": 1.6508899280515134e-08,
      "loss": 1.76,
      "step": 1460
    },
    {
      "epoch": 0.09612316397190651,
      "grad_norm": 0.5953572988510132,
      "learning_rate": 1.3872183137397799e-08,
      "loss": 1.6664,
      "step": 1461
    },
    {
      "epoch": 0.09618895669194204,
      "grad_norm": 0.6339824199676514,
      "learning_rate": 1.1464706174701967e-08,
      "loss": 1.5749,
      "step": 1462
    },
    {
      "epoch": 0.09625474941197756,
      "grad_norm": 0.598288357257843,
      "learning_rate": 9.286479433257e-09,
      "loss": 1.5851,
      "step": 1463
    },
    {
      "epoch": 0.09632054213201309,
      "grad_norm": 0.6078794002532959,
      "learning_rate": 7.337512902522159e-09,
      "loss": 1.6656,
      "step": 1464
    },
    {
      "epoch": 0.09638633485204862,
      "grad_norm": 0.6280359029769897,
      "learning_rate": 5.6178155205754975e-09,
      "loss": 1.6216,
      "step": 1465
    },
    {
      "epoch": 0.09645212757208414,
      "grad_norm": 0.5920078754425049,
      "learning_rate": 4.127395174036153e-09,
      "loss": 1.5143,
      "step": 1466
    },
    {
      "epoch": 0.09651792029211968,
      "grad_norm": 0.6147146224975586,
      "learning_rate": 2.866258698064339e-09,
      "loss": 1.559,
      "step": 1467
    },
    {
      "epoch": 0.09658371301215521,
      "grad_norm": 0.6601701974868774,
      "learning_rate": 1.8344118763002903e-09,
      "loss": 1.593,
      "step": 1468
    },
    {
      "epoch": 0.09664950573219074,
      "grad_norm": 0.6434551477432251,
      "learning_rate": 1.0318594408476045e-09,
      "loss": 1.5645,
      "step": 1469
    },
    {
      "epoch": 0.09671529845222626,
      "grad_norm": 0.6171491146087646,
      "learning_rate": 4.586050722621416e-10,
      "loss": 1.589,
      "step": 1470
    },
    {
      "epoch": 0.09678109117226179,
      "grad_norm": 0.6035428047180176,
      "learning_rate": 1.1465139951316595e-10,
      "loss": 1.5367,
      "step": 1471
    },
    {
      "epoch": 0.09684688389229731,
      "grad_norm": 0.6608728170394897,
      "learning_rate": 0.0,
      "loss": 1.517,
      "step": 1472
    }
  ],
  "logging_steps": 1,
  "max_steps": 1472,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 368,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.9072999168294257e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}