{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 313,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0032,
      "grad_norm": 1.2938580513000488,
      "learning_rate": 0.0,
      "loss": 2.3978,
      "step": 1
    },
    {
      "epoch": 0.0064,
      "grad_norm": 1.3802108764648438,
      "learning_rate": 1.5625e-06,
      "loss": 2.1638,
      "step": 2
    },
    {
      "epoch": 0.0096,
      "grad_norm": 1.401850700378418,
      "learning_rate": 3.125e-06,
      "loss": 2.0659,
      "step": 3
    },
    {
      "epoch": 0.0128,
      "grad_norm": NaN,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 2.466,
      "step": 4
    },
    {
      "epoch": 0.016,
      "grad_norm": 1.5032564401626587,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 2.088,
      "step": 5
    },
    {
      "epoch": 0.0192,
      "grad_norm": 3.115506172180176,
      "learning_rate": 6.25e-06,
      "loss": 2.0598,
      "step": 6
    },
    {
      "epoch": 0.0224,
      "grad_norm": 1.3768525123596191,
      "learning_rate": 7.8125e-06,
      "loss": 2.2994,
      "step": 7
    },
    {
      "epoch": 0.0256,
      "grad_norm": 1.4846614599227905,
      "learning_rate": 9.375000000000001e-06,
      "loss": 2.0359,
      "step": 8
    },
    {
      "epoch": 0.0288,
      "grad_norm": 4.242325782775879,
      "learning_rate": 1.09375e-05,
      "loss": 1.9999,
      "step": 9
    },
    {
      "epoch": 0.032,
      "grad_norm": 2.884045362472534,
      "learning_rate": 1.25e-05,
      "loss": 1.6543,
      "step": 10
    },
    {
      "epoch": 0.0352,
      "grad_norm": 1.1325325965881348,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 2.5745,
      "step": 11
    },
    {
      "epoch": 0.0384,
      "grad_norm": 1.1092923879623413,
      "learning_rate": 1.5625e-05,
      "loss": 2.3241,
      "step": 12
    },
    {
      "epoch": 0.0416,
      "grad_norm": 1.071656346321106,
      "learning_rate": 1.71875e-05,
      "loss": 2.3016,
      "step": 13
    },
    {
      "epoch": 0.0448,
      "grad_norm": 1.2169474363327026,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 1.9022,
      "step": 14
    },
    {
      "epoch": 0.048,
      "grad_norm": 1.1128144264221191,
      "learning_rate": 2.0312500000000002e-05,
      "loss": 2.0983,
      "step": 15
    },
    {
      "epoch": 0.0512,
      "grad_norm": 1.367555022239685,
      "learning_rate": 2.1875e-05,
      "loss": 2.3887,
      "step": 16
    },
    {
      "epoch": 0.0544,
      "grad_norm": 1.3732932806015015,
      "learning_rate": 2.34375e-05,
      "loss": 2.4338,
      "step": 17
    },
    {
      "epoch": 0.0576,
      "grad_norm": 1.0335572957992554,
      "learning_rate": 2.5e-05,
      "loss": 2.1441,
      "step": 18
    },
    {
      "epoch": 0.0608,
      "grad_norm": 1.057332158088684,
      "learning_rate": 2.6562500000000002e-05,
      "loss": 2.4553,
      "step": 19
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.9568037986755371,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 2.1104,
      "step": 20
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.9306254982948303,
      "learning_rate": 2.96875e-05,
      "loss": 1.936,
      "step": 21
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.9852834343910217,
      "learning_rate": 3.125e-05,
      "loss": 2.3996,
      "step": 22
    },
    {
      "epoch": 0.0736,
      "grad_norm": 1.0083969831466675,
      "learning_rate": 3.2812500000000005e-05,
      "loss": 2.3265,
      "step": 23
    },
    {
      "epoch": 0.0768,
      "grad_norm": 2.5204765796661377,
      "learning_rate": 3.4375e-05,
      "loss": 2.3637,
      "step": 24
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.87235027551651,
      "learning_rate": 3.59375e-05,
      "loss": 2.049,
      "step": 25
    },
    {
      "epoch": 0.0832,
      "grad_norm": 1.0738499164581299,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.2154,
      "step": 26
    },
    {
      "epoch": 0.0864,
      "grad_norm": 1.1177905797958374,
      "learning_rate": 3.90625e-05,
      "loss": 2.2278,
      "step": 27
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.9642391800880432,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 2.1245,
      "step": 28
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.9148507118225098,
      "learning_rate": 4.21875e-05,
      "loss": 2.3712,
      "step": 29
    },
    {
      "epoch": 0.096,
      "grad_norm": 1.0361849069595337,
      "learning_rate": 4.375e-05,
      "loss": 2.1405,
      "step": 30
    },
    {
      "epoch": 0.0992,
      "grad_norm": 7.4032697677612305,
      "learning_rate": 4.5312500000000004e-05,
      "loss": 1.8717,
      "step": 31
    },
    {
      "epoch": 0.1024,
      "grad_norm": 1.011413812637329,
      "learning_rate": 4.6875e-05,
      "loss": 2.0154,
      "step": 32
    },
    {
      "epoch": 0.1056,
      "grad_norm": 1.0620793104171753,
      "learning_rate": 4.8437500000000005e-05,
      "loss": 2.2802,
      "step": 33
    },
    {
      "epoch": 0.1088,
      "grad_norm": 1.0524107217788696,
      "learning_rate": 5e-05,
      "loss": 2.1007,
      "step": 34
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.9986498355865479,
      "learning_rate": 4.999843759868819e-05,
      "loss": 2.0176,
      "step": 35
    },
    {
      "epoch": 0.1152,
      "grad_norm": 1.047672986984253,
      "learning_rate": 4.9993750590040575e-05,
      "loss": 2.3632,
      "step": 36
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.9666229486465454,
      "learning_rate": 4.998593955989626e-05,
      "loss": 2.0895,
      "step": 37
    },
    {
      "epoch": 0.1216,
      "grad_norm": 1.1995681524276733,
      "learning_rate": 4.9975005484572305e-05,
      "loss": 1.9112,
      "step": 38
    },
    {
      "epoch": 0.1248,
      "grad_norm": 3.1392664909362793,
      "learning_rate": 4.996094973074183e-05,
      "loss": 2.0765,
      "step": 39
    },
    {
      "epoch": 0.128,
      "grad_norm": 1.1716688871383667,
      "learning_rate": 4.994377405526308e-05,
      "loss": 1.7757,
      "step": 40
    },
    {
      "epoch": 0.1312,
      "grad_norm": 0.9636698365211487,
      "learning_rate": 4.992348060495989e-05,
      "loss": 2.1896,
      "step": 41
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.9387962222099304,
      "learning_rate": 4.990007191635334e-05,
      "loss": 2.1875,
      "step": 42
    },
    {
      "epoch": 0.1376,
      "grad_norm": 0.918551504611969,
      "learning_rate": 4.987355091534468e-05,
      "loss": 2.1049,
      "step": 43
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.9996175169944763,
      "learning_rate": 4.9843920916849645e-05,
      "loss": 2.4059,
      "step": 44
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.9736499190330505,
      "learning_rate": 4.981118562438414e-05,
      "loss": 1.999,
      "step": 45
    },
    {
      "epoch": 0.1472,
      "grad_norm": 1.036577582359314,
      "learning_rate": 4.9775349129601243e-05,
      "loss": 2.0207,
      "step": 46
    },
    {
      "epoch": 0.1504,
      "grad_norm": 0.9663489460945129,
      "learning_rate": 4.973641591177991e-05,
      "loss": 1.8309,
      "step": 47
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.9678478837013245,
      "learning_rate": 4.969439083726496e-05,
      "loss": 1.8601,
      "step": 48
    },
    {
      "epoch": 0.1568,
      "grad_norm": 1.1323456764221191,
      "learning_rate": 4.964927915885893e-05,
      "loss": 1.9505,
      "step": 49
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8682395219802856,
      "learning_rate": 4.960108651516545e-05,
      "loss": 1.7868,
      "step": 50
    },
    {
      "epoch": 0.1632,
      "grad_norm": 0.9288162589073181,
      "learning_rate": 4.954981892988451e-05,
      "loss": 2.1843,
      "step": 51
    },
    {
      "epoch": 0.1664,
      "grad_norm": 1.0266203880310059,
      "learning_rate": 4.949548281105951e-05,
      "loss": 2.3133,
      "step": 52
    },
    {
      "epoch": 0.1696,
      "grad_norm": 0.9135573506355286,
      "learning_rate": 4.943808495027631e-05,
      "loss": 2.1877,
      "step": 53
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.942903995513916,
      "learning_rate": 4.937763252181434e-05,
      "loss": 1.788,
      "step": 54
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.918860137462616,
      "learning_rate": 4.93141330817499e-05,
      "loss": 2.2997,
      "step": 55
    },
    {
      "epoch": 0.1792,
      "grad_norm": 1.0075480937957764,
      "learning_rate": 4.924759456701167e-05,
      "loss": 2.2172,
      "step": 56
    },
    {
      "epoch": 0.1824,
      "grad_norm": 0.8399456143379211,
      "learning_rate": 4.917802529438864e-05,
      "loss": 1.6652,
      "step": 57
    },
    {
      "epoch": 0.1856,
      "grad_norm": 0.8925999402999878,
      "learning_rate": 4.910543395949067e-05,
      "loss": 1.8684,
      "step": 58
    },
    {
      "epoch": 0.1888,
      "grad_norm": 0.8900786638259888,
      "learning_rate": 4.9029829635661475e-05,
      "loss": 2.1491,
      "step": 59
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.9098831415176392,
      "learning_rate": 4.895122177284465e-05,
      "loss": 1.9692,
      "step": 60
    },
    {
      "epoch": 0.1952,
      "grad_norm": 0.9300256371498108,
      "learning_rate": 4.8869620196402436e-05,
      "loss": 2.1417,
      "step": 61
    },
    {
      "epoch": 0.1984,
      "grad_norm": 0.8800130486488342,
      "learning_rate": 4.878503510588765e-05,
      "loss": 2.1419,
      "step": 62
    },
    {
      "epoch": 0.2016,
      "grad_norm": 0.869717538356781,
      "learning_rate": 4.8697477073768766e-05,
      "loss": 2.0039,
      "step": 63
    },
    {
      "epoch": 0.2048,
      "grad_norm": 1.0832667350769043,
      "learning_rate": 4.8606957044108556e-05,
      "loss": 1.9359,
      "step": 64
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.8442544341087341,
      "learning_rate": 4.851348633119606e-05,
      "loss": 1.771,
      "step": 65
    },
    {
      "epoch": 0.2112,
      "grad_norm": 0.8307965397834778,
      "learning_rate": 4.8417076618132426e-05,
      "loss": 1.7329,
      "step": 66
    },
    {
      "epoch": 0.2144,
      "grad_norm": 0.8500248789787292,
      "learning_rate": 4.8317739955370636e-05,
      "loss": 1.927,
      "step": 67
    },
    {
      "epoch": 0.2176,
      "grad_norm": 0.8779441714286804,
      "learning_rate": 4.821548875920927e-05,
      "loss": 1.8811,
      "step": 68
    },
    {
      "epoch": 0.2208,
      "grad_norm": 0.9491339325904846,
      "learning_rate": 4.811033581024056e-05,
      "loss": 2.0995,
      "step": 69
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.9209153056144714,
      "learning_rate": 4.800229425175294e-05,
      "loss": 2.0308,
      "step": 70
    },
    {
      "epoch": 0.2272,
      "grad_norm": 0.9706084132194519,
      "learning_rate": 4.7891377588088223e-05,
      "loss": 2.1603,
      "step": 71
    },
    {
      "epoch": 0.2304,
      "grad_norm": 0.8401190638542175,
      "learning_rate": 4.777759968295369e-05,
      "loss": 1.8921,
      "step": 72
    },
    {
      "epoch": 0.2336,
      "grad_norm": 1.7388620376586914,
      "learning_rate": 4.766097475768919e-05,
      "loss": 1.9469,
      "step": 73
    },
    {
      "epoch": 0.2368,
      "grad_norm": 0.8813150525093079,
      "learning_rate": 4.754151738948962e-05,
      "loss": 1.8235,
      "step": 74
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8916892409324646,
      "learning_rate": 4.741924250958289e-05,
      "loss": 1.9513,
      "step": 75
    },
    {
      "epoch": 0.2432,
      "grad_norm": 0.8717350959777832,
      "learning_rate": 4.729416540136361e-05,
      "loss": 1.9231,
      "step": 76
    },
    {
      "epoch": 0.2464,
      "grad_norm": 0.9412696957588196,
      "learning_rate": 4.7166301698482815e-05,
      "loss": 2.099,
      "step": 77
    },
    {
      "epoch": 0.2496,
      "grad_norm": 0.9612340331077576,
      "learning_rate": 4.703566738289389e-05,
      "loss": 2.059,
      "step": 78
    },
    {
      "epoch": 0.2528,
      "grad_norm": 1.0139132738113403,
      "learning_rate": 4.69022787828549e-05,
      "loss": 1.6535,
      "step": 79
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.9509197473526001,
      "learning_rate": 4.676615257088776e-05,
      "loss": 2.1214,
      "step": 80
    },
    {
      "epoch": 0.2592,
      "grad_norm": 0.8892962336540222,
      "learning_rate": 4.662730576169423e-05,
      "loss": 2.1229,
      "step": 81
    },
    {
      "epoch": 0.2624,
      "grad_norm": 0.8373662829399109,
      "learning_rate": 4.6485755710029256e-05,
      "loss": 1.8558,
      "step": 82
    },
    {
      "epoch": 0.2656,
      "grad_norm": 0.8412055373191833,
      "learning_rate": 4.6341520108531746e-05,
      "loss": 1.7079,
      "step": 83
    },
    {
      "epoch": 0.2688,
      "grad_norm": 1.3762493133544922,
      "learning_rate": 4.619461698551315e-05,
      "loss": 1.8366,
      "step": 84
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.9610726237297058,
      "learning_rate": 4.604506470270403e-05,
      "loss": 2.1713,
      "step": 85
    },
    {
      "epoch": 0.2752,
      "grad_norm": 0.846570611000061,
      "learning_rate": 4.589288195295901e-05,
      "loss": 1.8648,
      "step": 86
    },
    {
      "epoch": 0.2784,
      "grad_norm": 0.9135278463363647,
      "learning_rate": 4.573808775792033e-05,
      "loss": 2.182,
      "step": 87
    },
    {
      "epoch": 0.2816,
      "grad_norm": 0.8865692615509033,
      "learning_rate": 4.5580701465640254e-05,
      "loss": 1.9423,
      "step": 88
    },
    {
      "epoch": 0.2848,
      "grad_norm": 0.8762995004653931,
      "learning_rate": 4.5420742748162734e-05,
      "loss": 1.9934,
      "step": 89
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.9263061881065369,
      "learning_rate": 4.525823159906459e-05,
      "loss": 2.2362,
      "step": 90
    },
    {
      "epoch": 0.2912,
      "grad_norm": 0.8523189425468445,
      "learning_rate": 4.509318833095642e-05,
      "loss": 2.1758,
      "step": 91
    },
    {
      "epoch": 0.2944,
      "grad_norm": 0.9841741919517517,
      "learning_rate": 4.492563357294369e-05,
      "loss": 2.0549,
      "step": 92
    },
    {
      "epoch": 0.2976,
      "grad_norm": 0.9290546178817749,
      "learning_rate": 4.475558826804833e-05,
      "loss": 1.8732,
      "step": 93
    },
    {
      "epoch": 0.3008,
      "grad_norm": 0.8647733926773071,
      "learning_rate": 4.458307367059092e-05,
      "loss": 2.0057,
      "step": 94
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.9161149859428406,
      "learning_rate": 4.440811134353412e-05,
      "loss": 1.7013,
      "step": 95
    },
    {
      "epoch": 0.3072,
      "grad_norm": 0.8401817083358765,
      "learning_rate": 4.42307231557875e-05,
      "loss": 2.0538,
      "step": 96
    },
    {
      "epoch": 0.3104,
      "grad_norm": 1.011531114578247,
      "learning_rate": 4.4050931279474015e-05,
      "loss": 1.998,
      "step": 97
    },
    {
      "epoch": 0.3136,
      "grad_norm": 0.8913917541503906,
      "learning_rate": 4.386875818715874e-05,
      "loss": 1.9491,
      "step": 98
    },
    {
      "epoch": 0.3168,
      "grad_norm": 0.9900378584861755,
      "learning_rate": 4.368422664903997e-05,
      "loss": 1.8276,
      "step": 99
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8170567750930786,
      "learning_rate": 4.349735973010305e-05,
      "loss": 1.8383,
      "step": 100
    },
    {
      "epoch": 0.3232,
      "grad_norm": 0.8767079710960388,
      "learning_rate": 4.330818078723755e-05,
      "loss": 1.8861,
      "step": 101
    },
    {
      "epoch": 0.3264,
      "grad_norm": 0.9571582078933716,
      "learning_rate": 4.311671346631774e-05,
      "loss": 1.8612,
      "step": 102
    },
    {
      "epoch": 0.3296,
      "grad_norm": 0.9190142154693604,
      "learning_rate": 4.292298169924709e-05,
      "loss": 2.1069,
      "step": 103
    },
    {
      "epoch": 0.3328,
      "grad_norm": 0.9096384644508362,
      "learning_rate": 4.272700970096696e-05,
      "loss": 2.0704,
      "step": 104
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.929742693901062,
      "learning_rate": 4.252882196642992e-05,
      "loss": 1.8269,
      "step": 105
    },
    {
      "epoch": 0.3392,
      "grad_norm": 0.8830273747444153,
      "learning_rate": 4.23284432675381e-05,
      "loss": 1.8447,
      "step": 106
    },
    {
      "epoch": 0.3424,
      "grad_norm": 0.866507887840271,
      "learning_rate": 4.212589865004684e-05,
      "loss": 2.2203,
      "step": 107
    },
    {
      "epoch": 0.3456,
      "grad_norm": 0.860091507434845,
      "learning_rate": 4.192121343043424e-05,
      "loss": 2.0754,
      "step": 108
    },
    {
      "epoch": 0.3488,
      "grad_norm": 0.9852222204208374,
      "learning_rate": 4.1714413192736754e-05,
      "loss": 1.7543,
      "step": 109
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.8804312944412231,
      "learning_rate": 4.150552378535137e-05,
      "loss": 2.0036,
      "step": 110
    },
    {
      "epoch": 0.3552,
      "grad_norm": 0.7997055649757385,
      "learning_rate": 4.1294571317804854e-05,
      "loss": 1.6802,
      "step": 111
    },
    {
      "epoch": 0.3584,
      "grad_norm": 0.9458661079406738,
      "learning_rate": 4.108158215749014e-05,
      "loss": 2.1944,
      "step": 112
    },
    {
      "epoch": 0.3616,
      "grad_norm": 0.9290521144866943,
      "learning_rate": 4.0866582926370725e-05,
      "loss": 1.9413,
      "step": 113
    },
    {
      "epoch": 0.3648,
      "grad_norm": 0.9188751578330994,
      "learning_rate": 4.064960049765304e-05,
      "loss": 2.1782,
      "step": 114
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.9197388887405396,
      "learning_rate": 4.043066199242762e-05,
      "loss": 2.301,
      "step": 115
    },
    {
      "epoch": 0.3712,
      "grad_norm": 0.8998375535011292,
      "learning_rate": 4.020979477627907e-05,
      "loss": 1.8035,
      "step": 116
    },
    {
      "epoch": 0.3744,
      "grad_norm": 0.8178901672363281,
      "learning_rate": 3.998702645586565e-05,
      "loss": 1.8764,
      "step": 117
    },
    {
      "epoch": 0.3776,
      "grad_norm": 0.9451196789741516,
      "learning_rate": 3.976238487546864e-05,
      "loss": 1.8488,
      "step": 118
    },
    {
      "epoch": 0.3808,
      "grad_norm": 0.8145877718925476,
      "learning_rate": 3.953589811351204e-05,
      "loss": 1.8894,
      "step": 119
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.8082287311553955,
      "learning_rate": 3.930759447905298e-05,
      "loss": 1.7236,
      "step": 120
    },
    {
      "epoch": 0.3872,
      "grad_norm": 1.0476171970367432,
      "learning_rate": 3.907750250824327e-05,
      "loss": 2.1334,
      "step": 121
    },
    {
      "epoch": 0.3904,
      "grad_norm": 0.9171006679534912,
      "learning_rate": 3.884565096076269e-05,
      "loss": 1.7087,
      "step": 122
    },
    {
      "epoch": 0.3936,
      "grad_norm": 0.9148180484771729,
      "learning_rate": 3.861206881622419e-05,
      "loss": 2.1576,
      "step": 123
    },
    {
      "epoch": 0.3968,
      "grad_norm": 0.8305053114891052,
      "learning_rate": 3.837678527055168e-05,
      "loss": 1.5619,
      "step": 124
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8816688060760498,
      "learning_rate": 3.813982973233083e-05,
      "loss": 1.7811,
      "step": 125
    },
    {
      "epoch": 0.4032,
      "grad_norm": 0.833396315574646,
      "learning_rate": 3.7901231819133105e-05,
      "loss": 1.9271,
      "step": 126
    },
    {
      "epoch": 0.4064,
      "grad_norm": 0.9301218390464783,
      "learning_rate": 3.766102135381393e-05,
      "loss": 2.1721,
      "step": 127
    },
    {
      "epoch": 0.4096,
      "grad_norm": 0.8964502215385437,
      "learning_rate": 3.741922836078499e-05,
      "loss": 1.7697,
      "step": 128
    },
    {
      "epoch": 0.4128,
      "grad_norm": 0.919535219669342,
      "learning_rate": 3.717588306226143e-05,
      "loss": 1.9075,
      "step": 129
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.9279409646987915,
      "learning_rate": 3.693101587448436e-05,
      "loss": 1.8996,
      "step": 130
    },
    {
      "epoch": 0.4192,
      "grad_norm": 0.8635279536247253,
      "learning_rate": 3.6684657403919005e-05,
      "loss": 1.8244,
      "step": 131
    },
    {
      "epoch": 0.4224,
      "grad_norm": 0.913209855556488,
      "learning_rate": 3.6436838443429175e-05,
      "loss": 2.0287,
      "step": 132
    },
    {
      "epoch": 0.4256,
      "grad_norm": 0.8230111002922058,
      "learning_rate": 3.618758996842839e-05,
      "loss": 1.8636,
      "step": 133
    },
    {
      "epoch": 0.4288,
      "grad_norm": 0.8902285099029541,
      "learning_rate": 3.5936943133008183e-05,
      "loss": 2.3189,
      "step": 134
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.8231277465820312,
      "learning_rate": 3.568492926604412e-05,
      "loss": 1.7345,
      "step": 135
    },
    {
      "epoch": 0.4352,
      "grad_norm": 0.82820063829422,
      "learning_rate": 3.5431579867279905e-05,
      "loss": 1.8564,
      "step": 136
    },
    {
      "epoch": 0.4384,
      "grad_norm": 0.8306258320808411,
      "learning_rate": 3.517692660339018e-05,
      "loss": 1.8671,
      "step": 137
    },
    {
      "epoch": 0.4416,
      "grad_norm": 0.8603615760803223,
      "learning_rate": 3.492100130402242e-05,
      "loss": 1.737,
      "step": 138
    },
    {
      "epoch": 0.4448,
      "grad_norm": 0.8179602026939392,
      "learning_rate": 3.4663835957818515e-05,
      "loss": 1.8263,
      "step": 139
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.8565822839736938,
      "learning_rate": 3.440546270841639e-05,
      "loss": 1.8216,
      "step": 140
    },
    {
      "epoch": 0.4512,
      "grad_norm": 0.9373151659965515,
      "learning_rate": 3.414591385043237e-05,
      "loss": 2.236,
      "step": 141
    },
    {
      "epoch": 0.4544,
      "grad_norm": 0.8852721452713013,
      "learning_rate": 3.3885221825424537e-05,
      "loss": 1.7058,
      "step": 142
    },
    {
      "epoch": 0.4576,
      "grad_norm": 0.8958799839019775,
      "learning_rate": 3.362341921783784e-05,
      "loss": 2.0029,
      "step": 143
    },
    {
      "epoch": 0.4608,
      "grad_norm": 0.9146959185600281,
      "learning_rate": 3.336053875093128e-05,
      "loss": 1.9745,
      "step": 144
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.8560377359390259,
      "learning_rate": 3.309661328268776e-05,
      "loss": 1.939,
      "step": 145
    },
    {
      "epoch": 0.4672,
      "grad_norm": 0.9292604327201843,
      "learning_rate": 3.283167580170712e-05,
      "loss": 2.1731,
      "step": 146
    },
    {
      "epoch": 0.4704,
      "grad_norm": 0.8372029066085815,
      "learning_rate": 3.256575942308278e-05,
      "loss": 1.7536,
      "step": 147
    },
    {
      "epoch": 0.4736,
      "grad_norm": 1.0057106018066406,
      "learning_rate": 3.229889738426264e-05,
      "loss": 2.0545,
      "step": 148
    },
    {
      "epoch": 0.4768,
      "grad_norm": 0.905636191368103,
      "learning_rate": 3.203112304089466e-05,
      "loss": 2.0688,
      "step": 149
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9915895462036133,
      "learning_rate": 3.176246986265767e-05,
      "loss": 1.7727,
      "step": 150
    },
    {
      "epoch": 0.4832,
      "grad_norm": 0.8286227583885193,
      "learning_rate": 3.149297142907792e-05,
      "loss": 1.9014,
      "step": 151
    },
    {
      "epoch": 0.4864,
      "grad_norm": 0.8335422873497009,
      "learning_rate": 3.122266142533191e-05,
      "loss": 1.8484,
      "step": 152
    },
    {
      "epoch": 0.4896,
      "grad_norm": 0.8963138461112976,
      "learning_rate": 3.095157363803598e-05,
      "loss": 2.042,
      "step": 153
    },
    {
      "epoch": 0.4928,
      "grad_norm": 0.9721676707267761,
      "learning_rate": 3.06797419510233e-05,
      "loss": 1.7216,
      "step": 154
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.927716076374054,
      "learning_rate": 3.0407200341108617e-05,
      "loss": 1.948,
      "step": 155
    },
    {
      "epoch": 0.4992,
      "grad_norm": 0.830467164516449,
      "learning_rate": 3.013398287384144e-05,
      "loss": 1.8576,
      "step": 156
    },
    {
      "epoch": 0.5024,
      "grad_norm": 0.9040769338607788,
      "learning_rate": 2.986012369924811e-05,
      "loss": 1.8908,
      "step": 157
    },
    {
      "epoch": 0.5056,
      "grad_norm": 0.8086222410202026,
      "learning_rate": 2.9585657047563315e-05,
      "loss": 1.868,
      "step": 158
    },
    {
      "epoch": 0.5088,
      "grad_norm": 0.8116671442985535,
      "learning_rate": 2.931061722495159e-05,
      "loss": 1.7886,
      "step": 159
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.9567749500274658,
      "learning_rate": 2.9035038609219306e-05,
      "loss": 1.853,
      "step": 160
    },
    {
      "epoch": 0.5152,
      "grad_norm": 0.9087838530540466,
      "learning_rate": 2.875895564551772e-05,
      "loss": 1.9346,
      "step": 161
    },
    {
      "epoch": 0.5184,
      "grad_norm": 1.079892873764038,
      "learning_rate": 2.8482402842037614e-05,
      "loss": 1.5922,
      "step": 162
    },
    {
      "epoch": 0.5216,
      "grad_norm": 0.8078391551971436,
      "learning_rate": 2.8205414765696003e-05,
      "loss": 1.6757,
      "step": 163
    },
    {
      "epoch": 0.5248,
      "grad_norm": 0.9625094532966614,
      "learning_rate": 2.792802603781562e-05,
      "loss": 2.0973,
      "step": 164
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.8933418393135071,
      "learning_rate": 2.7650271329797427e-05,
      "loss": 1.9035,
      "step": 165
    },
    {
      "epoch": 0.5312,
      "grad_norm": 0.972737193107605,
      "learning_rate": 2.737218535878705e-05,
      "loss": 1.694,
      "step": 166
    },
    {
      "epoch": 0.5344,
      "grad_norm": 0.8630124926567078,
      "learning_rate": 2.7093802883335357e-05,
      "loss": 1.9106,
      "step": 167
    },
    {
      "epoch": 0.5376,
      "grad_norm": 1.0283786058425903,
      "learning_rate": 2.6815158699053932e-05,
      "loss": 2.0379,
      "step": 168
    },
    {
      "epoch": 0.5408,
      "grad_norm": 0.8262351751327515,
      "learning_rate": 2.6536287634265918e-05,
      "loss": 1.7445,
      "step": 169
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.940129816532135,
      "learning_rate": 2.6257224545652688e-05,
      "loss": 2.1189,
      "step": 170
    },
    {
      "epoch": 0.5472,
      "grad_norm": 0.9087008833885193,
      "learning_rate": 2.5978004313897104e-05,
      "loss": 1.9498,
      "step": 171
    },
    {
      "epoch": 0.5504,
      "grad_norm": 0.8524187207221985,
      "learning_rate": 2.569866183932368e-05,
      "loss": 1.9588,
      "step": 172
    },
    {
      "epoch": 0.5536,
      "grad_norm": 0.8133301734924316,
      "learning_rate": 2.5419232037536316e-05,
      "loss": 1.8297,
      "step": 173
    },
    {
      "epoch": 0.5568,
      "grad_norm": 0.8447262644767761,
      "learning_rate": 2.5139749835054123e-05,
      "loss": 1.7685,
      "step": 174
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8986214995384216,
      "learning_rate": 2.4860250164945876e-05,
      "loss": 1.7743,
      "step": 175
    },
    {
      "epoch": 0.5632,
      "grad_norm": 0.9351289868354797,
      "learning_rate": 2.4580767962463687e-05,
      "loss": 1.8997,
      "step": 176
    },
    {
      "epoch": 0.5664,
      "grad_norm": 0.9037435054779053,
      "learning_rate": 2.4301338160676324e-05,
      "loss": 1.9691,
      "step": 177
    },
    {
      "epoch": 0.5696,
      "grad_norm": 0.849233865737915,
      "learning_rate": 2.40219956861029e-05,
      "loss": 1.8385,
      "step": 178
    },
    {
      "epoch": 0.5728,
      "grad_norm": 0.9789981245994568,
      "learning_rate": 2.374277545434732e-05,
      "loss": 1.8535,
      "step": 179
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.8241089582443237,
      "learning_rate": 2.346371236573409e-05,
      "loss": 1.6937,
      "step": 180
    },
    {
      "epoch": 0.5792,
      "grad_norm": 0.8240427374839783,
      "learning_rate": 2.318484130094607e-05,
      "loss": 1.8915,
      "step": 181
    },
    {
      "epoch": 0.5824,
      "grad_norm": 0.9362565875053406,
      "learning_rate": 2.2906197116664653e-05,
      "loss": 1.8763,
      "step": 182
    },
    {
      "epoch": 0.5856,
      "grad_norm": 0.880988359451294,
      "learning_rate": 2.262781464121296e-05,
      "loss": 2.0785,
      "step": 183
    },
    {
      "epoch": 0.5888,
      "grad_norm": 0.8171457648277283,
      "learning_rate": 2.2349728670202582e-05,
      "loss": 1.8087,
      "step": 184
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.9022923111915588,
      "learning_rate": 2.2071973962184384e-05,
      "loss": 1.9251,
      "step": 185
    },
    {
      "epoch": 0.5952,
      "grad_norm": 0.8642033338546753,
      "learning_rate": 2.1794585234303993e-05,
      "loss": 1.7433,
      "step": 186
    },
    {
      "epoch": 0.5984,
      "grad_norm": 0.9066776633262634,
      "learning_rate": 2.1517597157962392e-05,
      "loss": 1.7444,
      "step": 187
    },
    {
      "epoch": 0.6016,
      "grad_norm": 1.0174524784088135,
      "learning_rate": 2.124104435448228e-05,
      "loss": 1.9772,
      "step": 188
    },
    {
      "epoch": 0.6048,
      "grad_norm": 0.9552676677703857,
      "learning_rate": 2.0964961390780703e-05,
      "loss": 1.8762,
      "step": 189
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.9591497778892517,
      "learning_rate": 2.0689382775048418e-05,
      "loss": 1.9189,
      "step": 190
    },
    {
      "epoch": 0.6112,
      "grad_norm": 0.858933687210083,
      "learning_rate": 2.0414342952436694e-05,
      "loss": 1.8193,
      "step": 191
    },
    {
      "epoch": 0.6144,
      "grad_norm": 0.8183876872062683,
      "learning_rate": 2.0139876300751904e-05,
      "loss": 1.6413,
      "step": 192
    },
    {
      "epoch": 0.6176,
      "grad_norm": 0.8995866775512695,
      "learning_rate": 1.9866017126158574e-05,
      "loss": 2.0298,
      "step": 193
    },
    {
      "epoch": 0.6208,
      "grad_norm": 0.8391925692558289,
      "learning_rate": 1.9592799658891385e-05,
      "loss": 1.8856,
      "step": 194
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.8501824736595154,
      "learning_rate": 1.9320258048976702e-05,
      "loss": 1.8058,
      "step": 195
    },
    {
      "epoch": 0.6272,
      "grad_norm": 0.8973059058189392,
      "learning_rate": 1.904842636196402e-05,
      "loss": 2.0181,
      "step": 196
    },
    {
      "epoch": 0.6304,
      "grad_norm": 0.8554884195327759,
      "learning_rate": 1.8777338574668095e-05,
      "loss": 1.8791,
      "step": 197
    },
    {
      "epoch": 0.6336,
      "grad_norm": 0.9751168489456177,
      "learning_rate": 1.850702857092208e-05,
      "loss": 2.1171,
      "step": 198
    },
    {
      "epoch": 0.6368,
      "grad_norm": 0.8239455223083496,
      "learning_rate": 1.8237530137342335e-05,
      "loss": 1.9513,
      "step": 199
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8867501020431519,
      "learning_rate": 1.796887695910535e-05,
      "loss": 2.1136,
      "step": 200
    },
    {
      "epoch": 0.6432,
      "grad_norm": 0.8486935496330261,
      "learning_rate": 1.7701102615737368e-05,
      "loss": 1.9146,
      "step": 201
    },
    {
      "epoch": 0.6464,
      "grad_norm": 0.8988845348358154,
      "learning_rate": 1.7434240576917226e-05,
      "loss": 1.997,
      "step": 202
    },
    {
      "epoch": 0.6496,
      "grad_norm": 0.8450900912284851,
      "learning_rate": 1.7168324198292888e-05,
      "loss": 1.8575,
      "step": 203
    },
    {
      "epoch": 0.6528,
      "grad_norm": 0.9029667377471924,
      "learning_rate": 1.6903386717312236e-05,
      "loss": 1.9964,
      "step": 204
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.8760489821434021,
      "learning_rate": 1.6639461249068726e-05,
      "loss": 2.1632,
      "step": 205
    },
    {
      "epoch": 0.6592,
      "grad_norm": 0.8516858816146851,
      "learning_rate": 1.637658078216217e-05,
      "loss": 1.8699,
      "step": 206
    },
    {
      "epoch": 0.6624,
      "grad_norm": 0.8790733814239502,
      "learning_rate": 1.6114778174575473e-05,
      "loss": 1.6239,
      "step": 207
    },
    {
      "epoch": 0.6656,
      "grad_norm": 0.9619203805923462,
      "learning_rate": 1.585408614956763e-05,
      "loss": 2.3154,
      "step": 208
    },
    {
      "epoch": 0.6688,
      "grad_norm": 0.8987338542938232,
      "learning_rate": 1.559453729158361e-05,
      "loss": 1.7751,
      "step": 209
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.8577209711074829,
      "learning_rate": 1.5336164042181494e-05,
      "loss": 1.8087,
      "step": 210
    },
    {
      "epoch": 0.6752,
      "grad_norm": 0.8519135117530823,
      "learning_rate": 1.5078998695977586e-05,
      "loss": 1.8299,
      "step": 211
    },
    {
      "epoch": 0.6784,
      "grad_norm": 0.9367295503616333,
      "learning_rate": 1.482307339660983e-05,
      "loss": 2.177,
      "step": 212
    },
    {
      "epoch": 0.6816,
      "grad_norm": 0.9398857951164246,
      "learning_rate": 1.4568420132720106e-05,
      "loss": 2.1118,
      "step": 213
    },
    {
      "epoch": 0.6848,
      "grad_norm": 0.8866623640060425,
      "learning_rate": 1.4315070733955888e-05,
      "loss": 2.0044,
      "step": 214
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.9234758615493774,
      "learning_rate": 1.4063056866991826e-05,
      "loss": 1.9386,
      "step": 215
    },
    {
      "epoch": 0.6912,
      "grad_norm": 0.8017230033874512,
      "learning_rate": 1.381241003157162e-05,
      "loss": 1.6357,
      "step": 216
    },
    {
      "epoch": 0.6944,
      "grad_norm": 0.8828235864639282,
      "learning_rate": 1.3563161556570826e-05,
      "loss": 1.9193,
      "step": 217
    },
    {
      "epoch": 0.6976,
      "grad_norm": 0.8700492978096008,
      "learning_rate": 1.3315342596080996e-05,
      "loss": 1.6987,
      "step": 218
    },
    {
      "epoch": 0.7008,
      "grad_norm": 0.8625661134719849,
      "learning_rate": 1.3068984125515644e-05,
      "loss": 1.9628,
      "step": 219
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.8373908400535583,
      "learning_rate": 1.2824116937738579e-05,
      "loss": 1.664,
      "step": 220
    },
    {
      "epoch": 0.7072,
      "grad_norm": 0.8767737150192261,
      "learning_rate": 1.2580771639215027e-05,
      "loss": 2.0254,
      "step": 221
    },
    {
      "epoch": 0.7104,
      "grad_norm": 0.8676634430885315,
      "learning_rate": 1.2338978646186084e-05,
      "loss": 1.7836,
      "step": 222
    },
    {
      "epoch": 0.7136,
      "grad_norm": 0.8328086733818054,
      "learning_rate": 1.2098768180866895e-05,
      "loss": 1.7046,
      "step": 223
    },
    {
      "epoch": 0.7168,
      "grad_norm": 0.8744992613792419,
      "learning_rate": 1.1860170267669174e-05,
      "loss": 1.8117,
      "step": 224
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7836005687713623,
      "learning_rate": 1.1623214729448317e-05,
      "loss": 1.7427,
      "step": 225
    },
    {
      "epoch": 0.7232,
      "grad_norm": 0.7778053283691406,
      "learning_rate": 1.1387931183775822e-05,
      "loss": 1.6208,
      "step": 226
    },
    {
      "epoch": 0.7264,
      "grad_norm": 0.895233690738678,
      "learning_rate": 1.1154349039237322e-05,
      "loss": 1.834,
      "step": 227
    },
    {
      "epoch": 0.7296,
      "grad_norm": 0.864975094795227,
      "learning_rate": 1.0922497491756734e-05,
      "loss": 1.7659,
      "step": 228
    },
    {
      "epoch": 0.7328,
      "grad_norm": 0.9300910830497742,
      "learning_rate": 1.0692405520947028e-05,
      "loss": 2.0783,
      "step": 229
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.9012324810028076,
      "learning_rate": 1.0464101886487958e-05,
      "loss": 1.9243,
      "step": 230
    },
    {
      "epoch": 0.7392,
      "grad_norm": 0.8135029673576355,
      "learning_rate": 1.0237615124531363e-05,
      "loss": 1.7498,
      "step": 231
    },
    {
      "epoch": 0.7424,
      "grad_norm": 0.8468223214149475,
      "learning_rate": 1.0012973544134358e-05,
      "loss": 1.8409,
      "step": 232
    },
    {
      "epoch": 0.7456,
      "grad_norm": 0.8117995262145996,
      "learning_rate": 9.79020522372093e-06,
      "loss": 1.7893,
      "step": 233
    },
    {
      "epoch": 0.7488,
      "grad_norm": 0.8934926986694336,
      "learning_rate": 9.569338007572382e-06,
      "loss": 1.9382,
      "step": 234
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.8657869100570679,
      "learning_rate": 9.35039950234696e-06,
      "loss": 1.7477,
      "step": 235
    },
    {
      "epoch": 0.7552,
      "grad_norm": 0.871850311756134,
      "learning_rate": 9.133417073629289e-06,
      "loss": 1.9922,
      "step": 236
    },
    {
      "epoch": 0.7584,
      "grad_norm": 0.8558037877082825,
      "learning_rate": 8.918417842509867e-06,
      "loss": 1.8059,
      "step": 237
    },
    {
      "epoch": 0.7616,
      "grad_norm": 0.8291991353034973,
      "learning_rate": 8.705428682195155e-06,
      "loss": 1.8627,
      "step": 238
    },
    {
      "epoch": 0.7648,
      "grad_norm": 0.9084640741348267,
      "learning_rate": 8.494476214648626e-06,
      "loss": 1.9113,
      "step": 239
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.8608958125114441,
      "learning_rate": 8.285586807263254e-06,
      "loss": 1.937,
      "step": 240
    },
    {
      "epoch": 0.7712,
      "grad_norm": 0.8367571830749512,
      "learning_rate": 8.078786569565763e-06,
      "loss": 1.7382,
      "step": 241
    },
    {
      "epoch": 0.7744,
      "grad_norm": 0.9098989367485046,
      "learning_rate": 7.874101349953167e-06,
      "loss": 2.0156,
      "step": 242
    },
    {
      "epoch": 0.7776,
      "grad_norm": 0.8325051069259644,
      "learning_rate": 7.671556732461905e-06,
      "loss": 1.6789,
      "step": 243
    },
    {
      "epoch": 0.7808,
      "grad_norm": 0.855073869228363,
      "learning_rate": 7.471178033570081e-06,
      "loss": 2.0656,
      "step": 244
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.8020650148391724,
      "learning_rate": 7.272990299033045e-06,
      "loss": 1.6928,
      "step": 245
    },
    {
      "epoch": 0.7872,
      "grad_norm": 0.980696439743042,
      "learning_rate": 7.077018300752916e-06,
      "loss": 1.8311,
      "step": 246
    },
    {
      "epoch": 0.7904,
      "grad_norm": 0.8271718621253967,
      "learning_rate": 6.883286533682265e-06,
      "loss": 1.9569,
      "step": 247
    },
    {
      "epoch": 0.7936,
      "grad_norm": 0.9465168118476868,
      "learning_rate": 6.691819212762454e-06,
      "loss": 2.1162,
      "step": 248
    },
    {
      "epoch": 0.7968,
      "grad_norm": 0.878783643245697,
      "learning_rate": 6.502640269896953e-06,
      "loss": 1.7634,
      "step": 249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9123842120170593,
      "learning_rate": 6.3157733509600355e-06,
      "loss": 2.2862,
      "step": 250
    },
    {
      "epoch": 0.8032,
      "grad_norm": 0.8306463956832886,
      "learning_rate": 6.1312418128412565e-06,
      "loss": 1.7149,
      "step": 251
    },
    {
      "epoch": 0.8064,
      "grad_norm": 0.8042676448822021,
      "learning_rate": 5.949068720525991e-06,
      "loss": 1.5697,
      "step": 252
    },
    {
      "epoch": 0.8096,
      "grad_norm": 0.8570838570594788,
      "learning_rate": 5.769276844212501e-06,
      "loss": 1.9575,
      "step": 253
    },
    {
      "epoch": 0.8128,
      "grad_norm": 0.853945791721344,
      "learning_rate": 5.591888656465874e-06,
      "loss": 1.6021,
      "step": 254
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.7824327945709229,
      "learning_rate": 5.416926329409083e-06,
      "loss": 1.7469,
      "step": 255
    },
    {
      "epoch": 0.8192,
      "grad_norm": 0.9162058234214783,
      "learning_rate": 5.244411731951671e-06,
      "loss": 1.9214,
      "step": 256
    },
    {
      "epoch": 0.8224,
      "grad_norm": 0.8960239887237549,
      "learning_rate": 5.074366427056309e-06,
      "loss": 1.8902,
      "step": 257
    },
    {
      "epoch": 0.8256,
      "grad_norm": 0.8540034294128418,
      "learning_rate": 4.90681166904359e-06,
      "loss": 1.7701,
      "step": 258
    },
    {
      "epoch": 0.8288,
      "grad_norm": 0.8078952431678772,
      "learning_rate": 4.741768400935417e-06,
      "loss": 1.7476,
      "step": 259
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.8801387548446655,
      "learning_rate": 4.579257251837271e-06,
      "loss": 1.9394,
      "step": 260
    },
    {
      "epoch": 0.8352,
      "grad_norm": 0.9091523885726929,
      "learning_rate": 4.419298534359759e-06,
      "loss": 1.9172,
      "step": 261
    },
    {
      "epoch": 0.8384,
      "grad_norm": 0.86611407995224,
      "learning_rate": 4.261912242079674e-06,
      "loss": 1.7744,
      "step": 262
    },
    {
      "epoch": 0.8416,
      "grad_norm": 0.8813964128494263,
      "learning_rate": 4.107118047040995e-06,
      "loss": 1.964,
      "step": 263
    },
    {
      "epoch": 0.8448,
      "grad_norm": 1.0881954431533813,
      "learning_rate": 3.954935297295975e-06,
      "loss": 1.7736,
      "step": 264
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.8743950128555298,
      "learning_rate": 3.8053830144868547e-06,
      "loss": 1.9119,
      "step": 265
    },
    {
      "epoch": 0.8512,
      "grad_norm": 0.8601691126823425,
      "learning_rate": 3.6584798914682582e-06,
      "loss": 1.8711,
      "step": 266
    },
    {
      "epoch": 0.8544,
      "grad_norm": 0.884376585483551,
      "learning_rate": 3.514244289970753e-06,
      "loss": 2.1294,
      "step": 267
    },
    {
      "epoch": 0.8576,
      "grad_norm": 0.9268573522567749,
      "learning_rate": 3.3726942383057763e-06,
      "loss": 2.061,
      "step": 268
    },
    {
      "epoch": 0.8608,
      "grad_norm": 0.7765485644340515,
      "learning_rate": 3.233847429112244e-06,
      "loss": 1.5819,
      "step": 269
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.8231096267700195,
      "learning_rate": 3.0977212171451e-06,
      "loss": 1.8227,
      "step": 270
    },
    {
      "epoch": 0.8672,
      "grad_norm": 0.8997424244880676,
      "learning_rate": 2.9643326171061165e-06,
      "loss": 1.6759,
      "step": 271
    },
    {
      "epoch": 0.8704,
      "grad_norm": 0.8399364948272705,
      "learning_rate": 2.833698301517185e-06,
      "loss": 1.6793,
      "step": 272
    },
    {
      "epoch": 0.8736,
      "grad_norm": 0.920159101486206,
      "learning_rate": 2.7058345986363974e-06,
      "loss": 2.0572,
      "step": 273
    },
    {
      "epoch": 0.8768,
      "grad_norm": 0.9057918787002563,
      "learning_rate": 2.5807574904171155e-06,
      "loss": 2.2139,
      "step": 274
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8899713158607483,
      "learning_rate": 2.4584826105103764e-06,
      "loss": 1.9027,
      "step": 275
    },
    {
      "epoch": 0.8832,
      "grad_norm": 1.0192358493804932,
      "learning_rate": 2.3390252423108076e-06,
      "loss": 1.8376,
      "step": 276
    },
    {
      "epoch": 0.8864,
      "grad_norm": 0.8623155951499939,
      "learning_rate": 2.222400317046308e-06,
      "loss": 1.4057,
      "step": 277
    },
    {
      "epoch": 0.8896,
      "grad_norm": 0.9236856698989868,
      "learning_rate": 2.108622411911773e-06,
      "loss": 1.9201,
      "step": 278
    },
    {
      "epoch": 0.8928,
      "grad_norm": 0.9156610369682312,
      "learning_rate": 1.997705748247067e-06,
      "loss": 2.0085,
      "step": 279
    },
    {
      "epoch": 0.896,
      "grad_norm": 1.2603415250778198,
      "learning_rate": 1.8896641897594492e-06,
      "loss": 1.7741,
      "step": 280
    },
    {
      "epoch": 0.8992,
      "grad_norm": 0.8280951380729675,
      "learning_rate": 1.78451124079074e-06,
      "loss": 1.8905,
      "step": 281
    },
    {
      "epoch": 0.9024,
      "grad_norm": 0.8544325828552246,
      "learning_rate": 1.6822600446293636e-06,
      "loss": 1.9967,
      "step": 282
    },
    {
      "epoch": 0.9056,
      "grad_norm": 0.9109524488449097,
      "learning_rate": 1.5829233818675766e-06,
      "loss": 1.73,
      "step": 283
    },
    {
      "epoch": 0.9088,
      "grad_norm": 0.8589962720870972,
      "learning_rate": 1.486513668803946e-06,
      "loss": 1.827,
      "step": 284
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.8872967958450317,
      "learning_rate": 1.3930429558914494e-06,
      "loss": 1.6534,
      "step": 285
    },
    {
      "epoch": 0.9152,
      "grad_norm": 0.8933804631233215,
      "learning_rate": 1.3025229262312366e-06,
      "loss": 1.9079,
      "step": 286
    },
    {
      "epoch": 0.9184,
      "grad_norm": 0.8584591746330261,
      "learning_rate": 1.214964894112361e-06,
      "loss": 1.8498,
      "step": 287
    },
    {
      "epoch": 0.9216,
      "grad_norm": 0.8858172297477722,
      "learning_rate": 1.1303798035975643e-06,
      "loss": 1.7681,
      "step": 288
    },
    {
      "epoch": 0.9248,
      "grad_norm": 0.8581066131591797,
      "learning_rate": 1.0487782271553504e-06,
      "loss": 1.9958,
      "step": 289
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.8548694849014282,
      "learning_rate": 9.701703643385295e-07,
      "loss": 1.8107,
      "step": 290
    },
    {
      "epoch": 0.9312,
      "grad_norm": 0.8163594603538513,
      "learning_rate": 8.94566040509337e-07,
      "loss": 1.8072,
      "step": 291
    },
    {
      "epoch": 0.9344,
      "grad_norm": 0.9220604300498962,
      "learning_rate": 8.219747056113586e-07,
      "loss": 2.109,
      "step": 292
    },
    {
      "epoch": 0.9376,
      "grad_norm": 0.8807504773139954,
      "learning_rate": 7.524054329883346e-07,
      "loss": 1.6955,
      "step": 293
    },
    {
      "epoch": 0.9408,
      "grad_norm": 0.8602936863899231,
      "learning_rate": 6.858669182500971e-07,
      "loss": 1.849,
      "step": 294
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.8880540132522583,
      "learning_rate": 6.223674781856592e-07,
      "loss": 2.0819,
      "step": 295
    },
    {
      "epoch": 0.9472,
      "grad_norm": 0.915898859500885,
      "learning_rate": 5.619150497236992e-07,
      "loss": 2.2103,
      "step": 296
    },
    {
      "epoch": 0.9504,
      "grad_norm": 0.9282847046852112,
      "learning_rate": 5.045171889404954e-07,
      "loss": 2.1302,
      "step": 297
    },
    {
      "epoch": 0.9536,
      "grad_norm": 0.801628053188324,
      "learning_rate": 4.501810701154907e-07,
      "loss": 1.8129,
      "step": 298
    },
    {
      "epoch": 0.9568,
      "grad_norm": 0.8396210670471191,
      "learning_rate": 3.98913484834551e-07,
      "loss": 1.3778,
      "step": 299
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8347736597061157,
      "learning_rate": 3.507208411410778e-07,
      "loss": 1.7383,
      "step": 300
    },
    {
      "epoch": 0.9632,
      "grad_norm": 0.876707136631012,
      "learning_rate": 3.0560916273504325e-07,
      "loss": 1.8849,
      "step": 301
    },
    {
      "epoch": 0.9664,
      "grad_norm": 0.9329042434692383,
      "learning_rate": 2.635840882200924e-07,
      "loss": 1.8339,
      "step": 302
    },
    {
      "epoch": 0.9696,
      "grad_norm": 0.8642157912254333,
      "learning_rate": 2.246508703987543e-07,
      "loss": 1.916,
      "step": 303
    },
    {
      "epoch": 0.9728,
      "grad_norm": 0.9208475351333618,
      "learning_rate": 1.8881437561586722e-07,
      "loss": 1.7845,
      "step": 304
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.8702243566513062,
      "learning_rate": 1.5607908315035667e-07,
      "loss": 2.0131,
      "step": 305
    },
    {
      "epoch": 0.9792,
      "grad_norm": 0.879602313041687,
      "learning_rate": 1.264490846553279e-07,
      "loss": 1.9444,
      "step": 306
    },
    {
      "epoch": 0.9824,
      "grad_norm": 0.8321172595024109,
      "learning_rate": 9.992808364666373e-08,
      "loss": 1.7554,
      "step": 307
    },
    {
      "epoch": 0.9856,
      "grad_norm": 0.9140818119049072,
      "learning_rate": 7.651939504010885e-08,
      "loss": 1.9615,
      "step": 308
    },
    {
      "epoch": 0.9888,
      "grad_norm": 0.8076279163360596,
      "learning_rate": 5.622594473692067e-08,
      "loss": 1.6552,
      "step": 309
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.8180491328239441,
      "learning_rate": 3.90502692581729e-08,
      "loss": 1.8337,
      "step": 310
    },
    {
      "epoch": 0.9952,
      "grad_norm": 0.919771671295166,
      "learning_rate": 2.4994515427695374e-08,
      "loss": 1.7055,
      "step": 311
    },
    {
      "epoch": 0.9984,
      "grad_norm": 0.8689696192741394,
      "learning_rate": 1.4060440103746964e-08,
      "loss": 1.8763,
      "step": 312
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3404301404953003,
      "learning_rate": 6.249409959421803e-09,
      "loss": 1.8715,
      "step": 313
    }
  ],
  "logging_steps": 1,
  "max_steps": 313,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.23639789568e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}