{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9170105456212746,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004585052728106373,
      "grad_norm": 51.42570877075195,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 7.6786,
      "step": 1
    },
    {
      "epoch": 0.0009170105456212746,
      "grad_norm": 46.08848190307617,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 7.6155,
      "step": 2
    },
    {
      "epoch": 0.001375515818431912,
      "grad_norm": 20.57785987854004,
      "learning_rate": 4.545454545454546e-05,
      "loss": 6.8878,
      "step": 3
    },
    {
      "epoch": 0.0018340210912425492,
      "grad_norm": 15.308398246765137,
      "learning_rate": 6.060606060606061e-05,
      "loss": 6.2138,
      "step": 4
    },
    {
      "epoch": 0.0022925263640531865,
      "grad_norm": 8.584917068481445,
      "learning_rate": 7.575757575757576e-05,
      "loss": 5.6896,
      "step": 5
    },
    {
      "epoch": 0.002751031636863824,
      "grad_norm": 7.2267913818359375,
      "learning_rate": 9.090909090909092e-05,
      "loss": 5.294,
      "step": 6
    },
    {
      "epoch": 0.0032095369096744614,
      "grad_norm": 5.866622447967529,
      "learning_rate": 0.00010606060606060606,
      "loss": 5.0773,
      "step": 7
    },
    {
      "epoch": 0.0036680421824850985,
      "grad_norm": 4.189790725708008,
      "learning_rate": 0.00012121212121212122,
      "loss": 4.7521,
      "step": 8
    },
    {
      "epoch": 0.0041265474552957355,
      "grad_norm": 3.377293586730957,
      "learning_rate": 0.00013636363636363637,
      "loss": 4.6252,
      "step": 9
    },
    {
      "epoch": 0.004585052728106373,
      "grad_norm": 3.0264148712158203,
      "learning_rate": 0.00015151515151515152,
      "loss": 4.4131,
      "step": 10
    },
    {
      "epoch": 0.0050435580009170105,
      "grad_norm": 2.7406539916992188,
      "learning_rate": 0.00016666666666666666,
      "loss": 4.1975,
      "step": 11
    },
    {
      "epoch": 0.005502063273727648,
      "grad_norm": 2.7898213863372803,
      "learning_rate": 0.00018181818181818183,
      "loss": 4.1548,
      "step": 12
    },
    {
      "epoch": 0.005960568546538285,
      "grad_norm": 5.627849578857422,
      "learning_rate": 0.00019696969696969695,
      "loss": 4.0663,
      "step": 13
    },
    {
      "epoch": 0.006419073819348923,
      "grad_norm": 2.0357418060302734,
      "learning_rate": 0.00021212121212121213,
      "loss": 4.0078,
      "step": 14
    },
    {
      "epoch": 0.0068775790921595595,
      "grad_norm": 2.8825297355651855,
      "learning_rate": 0.00022727272727272727,
      "loss": 3.8895,
      "step": 15
    },
    {
      "epoch": 0.007336084364970197,
      "grad_norm": 2.6802010536193848,
      "learning_rate": 0.00024242424242424245,
      "loss": 3.7835,
      "step": 16
    },
    {
      "epoch": 0.007794589637780834,
      "grad_norm": 1.7662291526794434,
      "learning_rate": 0.00025757575757575756,
      "loss": 3.7091,
      "step": 17
    },
    {
      "epoch": 0.008253094910591471,
      "grad_norm": 1.801316738128662,
      "learning_rate": 0.00027272727272727274,
      "loss": 3.5952,
      "step": 18
    },
    {
      "epoch": 0.00871160018340211,
      "grad_norm": 2.6820998191833496,
      "learning_rate": 0.0002878787878787879,
      "loss": 3.5717,
      "step": 19
    },
    {
      "epoch": 0.009170105456212746,
      "grad_norm": 1.5427918434143066,
      "learning_rate": 0.00030303030303030303,
      "loss": 3.6179,
      "step": 20
    },
    {
      "epoch": 0.009628610729023384,
      "grad_norm": 1.045554280281067,
      "learning_rate": 0.0003181818181818182,
      "loss": 3.5603,
      "step": 21
    },
    {
      "epoch": 0.010087116001834021,
      "grad_norm": 1.1188470125198364,
      "learning_rate": 0.0003333333333333333,
      "loss": 3.3913,
      "step": 22
    },
    {
      "epoch": 0.01054562127464466,
      "grad_norm": 1.1570559740066528,
      "learning_rate": 0.0003484848484848485,
      "loss": 3.3188,
      "step": 23
    },
    {
      "epoch": 0.011004126547455296,
      "grad_norm": 1.82534921169281,
      "learning_rate": 0.00036363636363636367,
      "loss": 3.4169,
      "step": 24
    },
    {
      "epoch": 0.011462631820265932,
      "grad_norm": 1.1535125970840454,
      "learning_rate": 0.0003787878787878788,
      "loss": 3.3374,
      "step": 25
    },
    {
      "epoch": 0.01192113709307657,
      "grad_norm": 0.7577817440032959,
      "learning_rate": 0.0003939393939393939,
      "loss": 3.1797,
      "step": 26
    },
    {
      "epoch": 0.012379642365887207,
      "grad_norm": 1.1359751224517822,
      "learning_rate": 0.00040909090909090913,
      "loss": 3.3149,
      "step": 27
    },
    {
      "epoch": 0.012838147638697846,
      "grad_norm": 0.8975465297698975,
      "learning_rate": 0.00042424242424242425,
      "loss": 3.183,
      "step": 28
    },
    {
      "epoch": 0.013296652911508482,
      "grad_norm": 0.8764389753341675,
      "learning_rate": 0.0004393939393939394,
      "loss": 3.3108,
      "step": 29
    },
    {
      "epoch": 0.013755158184319119,
      "grad_norm": 1.1733673810958862,
      "learning_rate": 0.00045454545454545455,
      "loss": 3.1533,
      "step": 30
    },
    {
      "epoch": 0.014213663457129757,
      "grad_norm": 0.7226018905639648,
      "learning_rate": 0.0004696969696969697,
      "loss": 3.1588,
      "step": 31
    },
    {
      "epoch": 0.014672168729940394,
      "grad_norm": 1.0062379837036133,
      "learning_rate": 0.0004848484848484849,
      "loss": 3.206,
      "step": 32
    },
    {
      "epoch": 0.015130674002751032,
      "grad_norm": 0.7839850783348083,
      "learning_rate": 0.0005,
      "loss": 3.1491,
      "step": 33
    },
    {
      "epoch": 0.015589179275561669,
      "grad_norm": 0.6202083230018616,
      "learning_rate": 0.0005151515151515151,
      "loss": 3.134,
      "step": 34
    },
    {
      "epoch": 0.016047684548372305,
      "grad_norm": 0.6104652881622314,
      "learning_rate": 0.0005303030303030302,
      "loss": 3.0357,
      "step": 35
    },
    {
      "epoch": 0.016506189821182942,
      "grad_norm": 0.5120542645454407,
      "learning_rate": 0.0005454545454545455,
      "loss": 2.9926,
      "step": 36
    },
    {
      "epoch": 0.016964695093993582,
      "grad_norm": 0.7102159857749939,
      "learning_rate": 0.0005606060606060606,
      "loss": 3.0157,
      "step": 37
    },
    {
      "epoch": 0.01742320036680422,
      "grad_norm": 0.5368555188179016,
      "learning_rate": 0.0005757575757575758,
      "loss": 3.0457,
      "step": 38
    },
    {
      "epoch": 0.017881705639614855,
      "grad_norm": 0.5477105975151062,
      "learning_rate": 0.0005909090909090909,
      "loss": 3.0307,
      "step": 39
    },
    {
      "epoch": 0.018340210912425492,
      "grad_norm": 0.5955835580825806,
      "learning_rate": 0.0006060606060606061,
      "loss": 2.9938,
      "step": 40
    },
    {
      "epoch": 0.01879871618523613,
      "grad_norm": 0.4676532447338104,
      "learning_rate": 0.0006212121212121212,
      "loss": 2.8828,
      "step": 41
    },
    {
      "epoch": 0.01925722145804677,
      "grad_norm": 0.7964932322502136,
      "learning_rate": 0.0006363636363636364,
      "loss": 2.9658,
      "step": 42
    },
    {
      "epoch": 0.019715726730857405,
      "grad_norm": 0.5311492085456848,
      "learning_rate": 0.0006515151515151515,
      "loss": 3.0011,
      "step": 43
    },
    {
      "epoch": 0.020174232003668042,
      "grad_norm": 0.5138784646987915,
      "learning_rate": 0.0006666666666666666,
      "loss": 2.8396,
      "step": 44
    },
    {
      "epoch": 0.02063273727647868,
      "grad_norm": 0.45181548595428467,
      "learning_rate": 0.0006818181818181818,
      "loss": 2.8284,
      "step": 45
    },
    {
      "epoch": 0.02109124254928932,
      "grad_norm": 0.5258728861808777,
      "learning_rate": 0.000696969696969697,
      "loss": 2.9634,
      "step": 46
    },
    {
      "epoch": 0.021549747822099955,
      "grad_norm": 0.4551188349723816,
      "learning_rate": 0.0007121212121212122,
      "loss": 2.8192,
      "step": 47
    },
    {
      "epoch": 0.02200825309491059,
      "grad_norm": 0.49938520789146423,
      "learning_rate": 0.0007272727272727273,
      "loss": 2.8208,
      "step": 48
    },
    {
      "epoch": 0.02246675836772123,
      "grad_norm": 0.5412823557853699,
      "learning_rate": 0.0007424242424242425,
      "loss": 2.972,
      "step": 49
    },
    {
      "epoch": 0.022925263640531865,
      "grad_norm": 0.4923003911972046,
      "learning_rate": 0.0007575757575757576,
      "loss": 2.8552,
      "step": 50
    },
    {
      "epoch": 0.023383768913342505,
      "grad_norm": 0.4990726709365845,
      "learning_rate": 0.0007727272727272727,
      "loss": 2.8631,
      "step": 51
    },
    {
      "epoch": 0.02384227418615314,
      "grad_norm": 0.5378886461257935,
      "learning_rate": 0.0007878787878787878,
      "loss": 2.9407,
      "step": 52
    },
    {
      "epoch": 0.024300779458963778,
      "grad_norm": 0.5009343028068542,
      "learning_rate": 0.000803030303030303,
      "loss": 2.894,
      "step": 53
    },
    {
      "epoch": 0.024759284731774415,
      "grad_norm": 0.5120266675949097,
      "learning_rate": 0.0008181818181818183,
      "loss": 2.8319,
      "step": 54
    },
    {
      "epoch": 0.02521779000458505,
      "grad_norm": 0.4928242564201355,
      "learning_rate": 0.0008333333333333334,
      "loss": 2.793,
      "step": 55
    },
    {
      "epoch": 0.02567629527739569,
      "grad_norm": 0.47817882895469666,
      "learning_rate": 0.0008484848484848485,
      "loss": 2.8467,
      "step": 56
    },
    {
      "epoch": 0.026134800550206328,
      "grad_norm": 0.4681580662727356,
      "learning_rate": 0.0008636363636363636,
      "loss": 2.8115,
      "step": 57
    },
    {
      "epoch": 0.026593305823016965,
      "grad_norm": 0.46094441413879395,
      "learning_rate": 0.0008787878787878789,
      "loss": 2.8867,
      "step": 58
    },
    {
      "epoch": 0.0270518110958276,
      "grad_norm": 0.42099979519844055,
      "learning_rate": 0.000893939393939394,
      "loss": 2.7524,
      "step": 59
    },
    {
      "epoch": 0.027510316368638238,
      "grad_norm": 0.4791127145290375,
      "learning_rate": 0.0009090909090909091,
      "loss": 2.7938,
      "step": 60
    },
    {
      "epoch": 0.027968821641448878,
      "grad_norm": 0.4675314128398895,
      "learning_rate": 0.0009242424242424242,
      "loss": 2.7002,
      "step": 61
    },
    {
      "epoch": 0.028427326914259515,
      "grad_norm": 0.4684106409549713,
      "learning_rate": 0.0009393939393939394,
      "loss": 2.7258,
      "step": 62
    },
    {
      "epoch": 0.02888583218707015,
      "grad_norm": 0.44446536898612976,
      "learning_rate": 0.0009545454545454546,
      "loss": 2.7697,
      "step": 63
    },
    {
      "epoch": 0.029344337459880788,
      "grad_norm": 0.4739280939102173,
      "learning_rate": 0.0009696969696969698,
      "loss": 2.8362,
      "step": 64
    },
    {
      "epoch": 0.029802842732691424,
      "grad_norm": 0.4399973452091217,
      "learning_rate": 0.000984848484848485,
      "loss": 2.7045,
      "step": 65
    },
    {
      "epoch": 0.030261348005502064,
      "grad_norm": 0.5304760932922363,
      "learning_rate": 0.001,
      "loss": 2.7938,
      "step": 66
    },
    {
      "epoch": 0.0307198532783127,
      "grad_norm": 0.46170714497566223,
      "learning_rate": 0.0009999994484067654,
      "loss": 2.7308,
      "step": 67
    },
    {
      "epoch": 0.031178358551123338,
      "grad_norm": 0.4680909812450409,
      "learning_rate": 0.0009999977936282788,
      "loss": 2.6792,
      "step": 68
    },
    {
      "epoch": 0.03163686382393398,
      "grad_norm": 0.42740005254745483,
      "learning_rate": 0.0009999950356681913,
      "loss": 2.7373,
      "step": 69
    },
    {
      "epoch": 0.03209536909674461,
      "grad_norm": 0.4212319552898407,
      "learning_rate": 0.0009999911745325876,
      "loss": 2.6639,
      "step": 70
    },
    {
      "epoch": 0.03255387436955525,
      "grad_norm": 0.3858383893966675,
      "learning_rate": 0.0009999862102299873,
      "loss": 2.5847,
      "step": 71
    },
    {
      "epoch": 0.033012379642365884,
      "grad_norm": 0.4052165448665619,
      "learning_rate": 0.0009999801427713433,
      "loss": 2.7228,
      "step": 72
    },
    {
      "epoch": 0.033470884915176524,
      "grad_norm": 0.3899725675582886,
      "learning_rate": 0.0009999729721700424,
      "loss": 2.5757,
      "step": 73
    },
    {
      "epoch": 0.033929390187987164,
      "grad_norm": 0.39178648591041565,
      "learning_rate": 0.000999964698441906,
      "loss": 2.6423,
      "step": 74
    },
    {
      "epoch": 0.0343878954607978,
      "grad_norm": 0.424039751291275,
      "learning_rate": 0.0009999553216051892,
      "loss": 2.6814,
      "step": 75
    },
    {
      "epoch": 0.03484640073360844,
      "grad_norm": 0.4265524446964264,
      "learning_rate": 0.00099994484168058,
      "loss": 2.6398,
      "step": 76
    },
    {
      "epoch": 0.03530490600641907,
      "grad_norm": 0.489754855632782,
      "learning_rate": 0.0009999332586912019,
      "loss": 2.6477,
      "step": 77
    },
    {
      "epoch": 0.03576341127922971,
      "grad_norm": 0.4495154619216919,
      "learning_rate": 0.0009999205726626108,
      "loss": 2.6418,
      "step": 78
    },
    {
      "epoch": 0.03622191655204035,
      "grad_norm": 0.4232162833213806,
      "learning_rate": 0.000999906783622797,
      "loss": 2.6179,
      "step": 79
    },
    {
      "epoch": 0.036680421824850984,
      "grad_norm": 0.4193362593650818,
      "learning_rate": 0.0009998918916021842,
      "loss": 2.6338,
      "step": 80
    },
    {
      "epoch": 0.037138927097661624,
      "grad_norm": 0.41505712270736694,
      "learning_rate": 0.0009998758966336297,
      "loss": 2.5507,
      "step": 81
    },
    {
      "epoch": 0.03759743237047226,
      "grad_norm": 0.4106564223766327,
      "learning_rate": 0.0009998587987524242,
      "loss": 2.5743,
      "step": 82
    },
    {
      "epoch": 0.0380559376432829,
      "grad_norm": 0.3779209852218628,
      "learning_rate": 0.0009998405979962926,
      "loss": 2.5734,
      "step": 83
    },
    {
      "epoch": 0.03851444291609354,
      "grad_norm": 0.3932972848415375,
      "learning_rate": 0.000999821294405392,
      "loss": 2.5552,
      "step": 84
    },
    {
      "epoch": 0.03897294818890417,
      "grad_norm": 0.46022525429725647,
      "learning_rate": 0.0009998008880223134,
      "loss": 2.6269,
      "step": 85
    },
    {
      "epoch": 0.03943145346171481,
      "grad_norm": 0.4375913739204407,
      "learning_rate": 0.000999779378892081,
      "loss": 2.5725,
      "step": 86
    },
    {
      "epoch": 0.039889958734525444,
      "grad_norm": 0.44462448358535767,
      "learning_rate": 0.0009997567670621522,
      "loss": 2.4073,
      "step": 87
    },
    {
      "epoch": 0.040348464007336084,
      "grad_norm": 0.42988884449005127,
      "learning_rate": 0.0009997330525824165,
      "loss": 2.4752,
      "step": 88
    },
    {
      "epoch": 0.040806969280146724,
      "grad_norm": 0.39581653475761414,
      "learning_rate": 0.0009997082355051976,
      "loss": 2.4895,
      "step": 89
    },
    {
      "epoch": 0.04126547455295736,
      "grad_norm": 0.433271586894989,
      "learning_rate": 0.000999682315885251,
      "loss": 2.5552,
      "step": 90
    },
    {
      "epoch": 0.041723979825768,
      "grad_norm": 0.4483635425567627,
      "learning_rate": 0.0009996552937797645,
      "loss": 2.449,
      "step": 91
    },
    {
      "epoch": 0.04218248509857864,
      "grad_norm": 0.38804078102111816,
      "learning_rate": 0.0009996271692483596,
      "loss": 2.4062,
      "step": 92
    },
    {
      "epoch": 0.04264099037138927,
      "grad_norm": 0.394492506980896,
      "learning_rate": 0.0009995979423530893,
      "loss": 2.405,
      "step": 93
    },
    {
      "epoch": 0.04309949564419991,
      "grad_norm": 0.3855603337287903,
      "learning_rate": 0.000999567613158439,
      "loss": 2.4798,
      "step": 94
    },
    {
      "epoch": 0.04355800091701054,
      "grad_norm": 0.3809776306152344,
      "learning_rate": 0.0009995361817313263,
      "loss": 2.4713,
      "step": 95
    },
    {
      "epoch": 0.04401650618982118,
      "grad_norm": 0.3731100261211395,
      "learning_rate": 0.0009995036481411004,
      "loss": 2.5103,
      "step": 96
    },
    {
      "epoch": 0.044475011462631824,
      "grad_norm": 0.37449559569358826,
      "learning_rate": 0.0009994700124595429,
      "loss": 2.5227,
      "step": 97
    },
    {
      "epoch": 0.04493351673544246,
      "grad_norm": 0.36096298694610596,
      "learning_rate": 0.0009994352747608663,
      "loss": 2.4429,
      "step": 98
    },
    {
      "epoch": 0.0453920220082531,
      "grad_norm": 0.400537371635437,
      "learning_rate": 0.0009993994351217151,
      "loss": 2.4483,
      "step": 99
    },
    {
      "epoch": 0.04585052728106373,
      "grad_norm": 0.3791731894016266,
      "learning_rate": 0.000999362493621165,
      "loss": 2.4843,
      "step": 100
    },
    {
      "epoch": 0.04630903255387437,
      "grad_norm": 0.4136984348297119,
      "learning_rate": 0.0009993244503407226,
      "loss": 2.4466,
      "step": 101
    },
    {
      "epoch": 0.04676753782668501,
      "grad_norm": 0.35446974635124207,
      "learning_rate": 0.0009992853053643258,
      "loss": 2.3404,
      "step": 102
    },
    {
      "epoch": 0.04722604309949564,
      "grad_norm": 0.3787573575973511,
      "learning_rate": 0.0009992450587783426,
      "loss": 2.3696,
      "step": 103
    },
    {
      "epoch": 0.04768454837230628,
      "grad_norm": 0.3715519309043884,
      "learning_rate": 0.000999203710671572,
      "loss": 2.4225,
      "step": 104
    },
    {
      "epoch": 0.048143053645116916,
      "grad_norm": 0.3520829677581787,
      "learning_rate": 0.0009991612611352438,
      "loss": 2.3857,
      "step": 105
    },
    {
      "epoch": 0.048601558917927556,
      "grad_norm": 0.39413443207740784,
      "learning_rate": 0.0009991177102630173,
      "loss": 2.4346,
      "step": 106
    },
    {
      "epoch": 0.049060064190738197,
      "grad_norm": 0.3751104176044464,
      "learning_rate": 0.0009990730581509817,
      "loss": 2.3278,
      "step": 107
    },
    {
      "epoch": 0.04951856946354883,
      "grad_norm": 0.35860374569892883,
      "learning_rate": 0.0009990273048976566,
      "loss": 2.3044,
      "step": 108
    },
    {
      "epoch": 0.04997707473635947,
      "grad_norm": 0.36340489983558655,
      "learning_rate": 0.0009989804506039905,
      "loss": 2.3386,
      "step": 109
    },
    {
      "epoch": 0.0504355800091701,
      "grad_norm": 0.37005776166915894,
      "learning_rate": 0.0009989324953733614,
      "loss": 2.435,
      "step": 110
    },
    {
      "epoch": 0.05089408528198074,
      "grad_norm": 0.33100616931915283,
      "learning_rate": 0.0009988834393115766,
      "loss": 2.4468,
      "step": 111
    },
    {
      "epoch": 0.05135259055479138,
      "grad_norm": 0.3255566954612732,
      "learning_rate": 0.000998833282526872,
      "loss": 2.3897,
      "step": 112
    },
    {
      "epoch": 0.051811095827602016,
      "grad_norm": 0.3608061671257019,
      "learning_rate": 0.0009987820251299122,
      "loss": 2.3033,
      "step": 113
    },
    {
      "epoch": 0.052269601100412656,
      "grad_norm": 0.3241058886051178,
      "learning_rate": 0.00099872966723379,
      "loss": 2.3538,
      "step": 114
    },
    {
      "epoch": 0.05272810637322329,
      "grad_norm": 0.3164448142051697,
      "learning_rate": 0.0009986762089540266,
      "loss": 2.3712,
      "step": 115
    },
    {
      "epoch": 0.05318661164603393,
      "grad_norm": 0.3709750175476074,
      "learning_rate": 0.0009986216504085709,
      "loss": 2.3907,
      "step": 116
    },
    {
      "epoch": 0.05364511691884457,
      "grad_norm": 0.3390752077102661,
      "learning_rate": 0.0009985659917177991,
      "loss": 2.4565,
      "step": 117
    },
    {
      "epoch": 0.0541036221916552,
      "grad_norm": 0.33107927441596985,
      "learning_rate": 0.0009985092330045155,
      "loss": 2.3542,
      "step": 118
    },
    {
      "epoch": 0.05456212746446584,
      "grad_norm": 0.3388403058052063,
      "learning_rate": 0.0009984513743939508,
      "loss": 2.3913,
      "step": 119
    },
    {
      "epoch": 0.055020632737276476,
      "grad_norm": 0.33465588092803955,
      "learning_rate": 0.0009983924160137626,
      "loss": 2.3258,
      "step": 120
    },
    {
      "epoch": 0.055479138010087116,
      "grad_norm": 0.3149769902229309,
      "learning_rate": 0.000998332357994035,
      "loss": 2.3117,
      "step": 121
    },
    {
      "epoch": 0.055937643282897756,
      "grad_norm": 0.3305632770061493,
      "learning_rate": 0.0009982712004672786,
      "loss": 2.3142,
      "step": 122
    },
    {
      "epoch": 0.05639614855570839,
      "grad_norm": 0.3419068157672882,
      "learning_rate": 0.0009982089435684295,
      "loss": 2.3168,
      "step": 123
    },
    {
      "epoch": 0.05685465382851903,
      "grad_norm": 0.3263803720474243,
      "learning_rate": 0.0009981455874348499,
      "loss": 2.3743,
      "step": 124
    },
    {
      "epoch": 0.05731315910132966,
      "grad_norm": 0.3268042802810669,
      "learning_rate": 0.0009980811322063269,
      "loss": 2.309,
      "step": 125
    },
    {
      "epoch": 0.0577716643741403,
      "grad_norm": 0.33107897639274597,
      "learning_rate": 0.0009980155780250728,
      "loss": 2.3232,
      "step": 126
    },
    {
      "epoch": 0.05823016964695094,
      "grad_norm": 0.32414108514785767,
      "learning_rate": 0.0009979489250357243,
      "loss": 2.2857,
      "step": 127
    },
    {
      "epoch": 0.058688674919761576,
      "grad_norm": 0.3102959394454956,
      "learning_rate": 0.0009978811733853431,
      "loss": 2.2996,
      "step": 128
    },
    {
      "epoch": 0.059147180192572216,
      "grad_norm": 0.31760498881340027,
      "learning_rate": 0.0009978123232234147,
      "loss": 2.2933,
      "step": 129
    },
    {
      "epoch": 0.05960568546538285,
      "grad_norm": 0.33521297574043274,
      "learning_rate": 0.000997742374701848,
      "loss": 2.3032,
      "step": 130
    },
    {
      "epoch": 0.06006419073819349,
      "grad_norm": 0.29083895683288574,
      "learning_rate": 0.0009976713279749754,
      "loss": 2.1832,
      "step": 131
    },
    {
      "epoch": 0.06052269601100413,
      "grad_norm": 0.31527042388916016,
      "learning_rate": 0.0009975991831995528,
      "loss": 2.2575,
      "step": 132
    },
    {
      "epoch": 0.06098120128381476,
      "grad_norm": 0.3197566568851471,
      "learning_rate": 0.0009975259405347581,
      "loss": 2.2804,
      "step": 133
    },
    {
      "epoch": 0.0614397065566254,
      "grad_norm": 0.3270002603530884,
      "learning_rate": 0.0009974516001421926,
      "loss": 2.3213,
      "step": 134
    },
    {
      "epoch": 0.061898211829436035,
      "grad_norm": 0.35190293192863464,
      "learning_rate": 0.000997376162185878,
      "loss": 2.2729,
      "step": 135
    },
    {
      "epoch": 0.062356717102246675,
      "grad_norm": 0.31001579761505127,
      "learning_rate": 0.0009972996268322594,
      "loss": 2.302,
      "step": 136
    },
    {
      "epoch": 0.06281522237505731,
      "grad_norm": 0.3456107974052429,
      "learning_rate": 0.0009972219942502017,
      "loss": 2.3256,
      "step": 137
    },
    {
      "epoch": 0.06327372764786796,
      "grad_norm": 0.303508996963501,
      "learning_rate": 0.0009971432646109918,
      "loss": 2.28,
      "step": 138
    },
    {
      "epoch": 0.06373223292067859,
      "grad_norm": 0.30973562598228455,
      "learning_rate": 0.0009970634380883365,
      "loss": 2.2698,
      "step": 139
    },
    {
      "epoch": 0.06419073819348922,
      "grad_norm": 0.30384132266044617,
      "learning_rate": 0.0009969825148583627,
      "loss": 2.2941,
      "step": 140
    },
    {
      "epoch": 0.06464924346629987,
      "grad_norm": 0.38114050030708313,
      "learning_rate": 0.0009969004950996173,
      "loss": 2.2323,
      "step": 141
    },
    {
      "epoch": 0.0651077487391105,
      "grad_norm": 0.31792521476745605,
      "learning_rate": 0.0009968173789930668,
      "loss": 2.3358,
      "step": 142
    },
    {
      "epoch": 0.06556625401192114,
      "grad_norm": 0.3239181339740753,
      "learning_rate": 0.0009967331667220958,
      "loss": 2.1962,
      "step": 143
    },
    {
      "epoch": 0.06602475928473177,
      "grad_norm": 0.3236413300037384,
      "learning_rate": 0.0009966478584725086,
      "loss": 2.2501,
      "step": 144
    },
    {
      "epoch": 0.06648326455754242,
      "grad_norm": 0.30811718106269836,
      "learning_rate": 0.0009965614544325263,
      "loss": 2.2778,
      "step": 145
    },
    {
      "epoch": 0.06694176983035305,
      "grad_norm": 0.29237669706344604,
      "learning_rate": 0.000996473954792789,
      "loss": 2.3162,
      "step": 146
    },
    {
      "epoch": 0.06740027510316368,
      "grad_norm": 0.3068889081478119,
      "learning_rate": 0.0009963853597463532,
      "loss": 2.2521,
      "step": 147
    },
    {
      "epoch": 0.06785878037597433,
      "grad_norm": 0.3155868649482727,
      "learning_rate": 0.000996295669488693,
      "loss": 2.2583,
      "step": 148
    },
    {
      "epoch": 0.06831728564878496,
      "grad_norm": 0.3209916949272156,
      "learning_rate": 0.0009962048842176979,
      "loss": 2.295,
      "step": 149
    },
    {
      "epoch": 0.0687757909215956,
      "grad_norm": 0.30611976981163025,
      "learning_rate": 0.0009961130041336748,
      "loss": 2.2545,
      "step": 150
    },
    {
      "epoch": 0.06923429619440624,
      "grad_norm": 0.30530020594596863,
      "learning_rate": 0.0009960200294393449,
      "loss": 2.3243,
      "step": 151
    },
    {
      "epoch": 0.06969280146721687,
      "grad_norm": 0.3141555190086365,
      "learning_rate": 0.0009959259603398453,
      "loss": 2.2995,
      "step": 152
    },
    {
      "epoch": 0.07015130674002751,
      "grad_norm": 0.3104286789894104,
      "learning_rate": 0.0009958307970427275,
      "loss": 2.1888,
      "step": 153
    },
    {
      "epoch": 0.07060981201283814,
      "grad_norm": 0.30111151933670044,
      "learning_rate": 0.0009957345397579572,
      "loss": 2.2673,
      "step": 154
    },
    {
      "epoch": 0.07106831728564879,
      "grad_norm": 0.307394802570343,
      "learning_rate": 0.0009956371886979138,
      "loss": 2.167,
      "step": 155
    },
    {
      "epoch": 0.07152682255845942,
      "grad_norm": 0.32675421237945557,
      "learning_rate": 0.00099553874407739,
      "loss": 2.3048,
      "step": 156
    },
    {
      "epoch": 0.07198532783127005,
      "grad_norm": 0.3122536540031433,
      "learning_rate": 0.0009954392061135916,
      "loss": 2.2605,
      "step": 157
    },
    {
      "epoch": 0.0724438331040807,
      "grad_norm": 0.30810093879699707,
      "learning_rate": 0.0009953385750261364,
      "loss": 2.2717,
      "step": 158
    },
    {
      "epoch": 0.07290233837689133,
      "grad_norm": 0.3372347950935364,
      "learning_rate": 0.0009952368510370538,
      "loss": 2.2379,
      "step": 159
    },
    {
      "epoch": 0.07336084364970197,
      "grad_norm": 0.30091676115989685,
      "learning_rate": 0.0009951340343707852,
      "loss": 2.2771,
      "step": 160
    },
    {
      "epoch": 0.07381934892251261,
      "grad_norm": 0.32447680830955505,
      "learning_rate": 0.0009950301252541823,
      "loss": 2.2585,
      "step": 161
    },
    {
      "epoch": 0.07427785419532325,
      "grad_norm": 0.31598275899887085,
      "learning_rate": 0.0009949251239165075,
      "loss": 2.185,
      "step": 162
    },
    {
      "epoch": 0.07473635946813388,
      "grad_norm": 0.28009629249572754,
      "learning_rate": 0.000994819030589433,
      "loss": 2.2906,
      "step": 163
    },
    {
      "epoch": 0.07519486474094451,
      "grad_norm": 0.30903327465057373,
      "learning_rate": 0.00099471184550704,
      "loss": 2.1986,
      "step": 164
    },
    {
      "epoch": 0.07565337001375516,
      "grad_norm": 0.3097755014896393,
      "learning_rate": 0.0009946035689058189,
      "loss": 2.2315,
      "step": 165
    },
    {
      "epoch": 0.0761118752865658,
      "grad_norm": 0.3114210069179535,
      "learning_rate": 0.0009944942010246681,
      "loss": 2.315,
      "step": 166
    },
    {
      "epoch": 0.07657038055937643,
      "grad_norm": 0.3382378816604614,
      "learning_rate": 0.0009943837421048942,
      "loss": 2.2577,
      "step": 167
    },
    {
      "epoch": 0.07702888583218707,
      "grad_norm": 0.31210529804229736,
      "learning_rate": 0.0009942721923902106,
      "loss": 2.1857,
      "step": 168
    },
    {
      "epoch": 0.07748739110499771,
      "grad_norm": 0.2818000316619873,
      "learning_rate": 0.0009941595521267377,
      "loss": 2.1294,
      "step": 169
    },
    {
      "epoch": 0.07794589637780834,
      "grad_norm": 0.30146366357803345,
      "learning_rate": 0.0009940458215630017,
      "loss": 2.2594,
      "step": 170
    },
    {
      "epoch": 0.07840440165061899,
      "grad_norm": 0.34887978434562683,
      "learning_rate": 0.0009939310009499348,
      "loss": 2.2829,
      "step": 171
    },
    {
      "epoch": 0.07886290692342962,
      "grad_norm": 0.3145126402378082,
      "learning_rate": 0.000993815090540874,
      "loss": 2.3282,
      "step": 172
    },
    {
      "epoch": 0.07932141219624025,
      "grad_norm": 0.3079446852207184,
      "learning_rate": 0.000993698090591561,
      "loss": 2.313,
      "step": 173
    },
    {
      "epoch": 0.07977991746905089,
      "grad_norm": 0.2939114570617676,
      "learning_rate": 0.0009935800013601416,
      "loss": 2.2192,
      "step": 174
    },
    {
      "epoch": 0.08023842274186153,
      "grad_norm": 0.28526419401168823,
      "learning_rate": 0.000993460823107164,
      "loss": 2.2355,
      "step": 175
    },
    {
      "epoch": 0.08069692801467217,
      "grad_norm": 0.26077085733413696,
      "learning_rate": 0.0009933405560955803,
      "loss": 2.2806,
      "step": 176
    },
    {
      "epoch": 0.0811554332874828,
      "grad_norm": 0.3125990331172943,
      "learning_rate": 0.0009932192005907446,
      "loss": 2.2727,
      "step": 177
    },
    {
      "epoch": 0.08161393856029345,
      "grad_norm": 0.2840162217617035,
      "learning_rate": 0.0009930967568604118,
      "loss": 2.3042,
      "step": 178
    },
    {
      "epoch": 0.08207244383310408,
      "grad_norm": 0.3122795820236206,
      "learning_rate": 0.000992973225174739,
      "loss": 2.2352,
      "step": 179
    },
    {
      "epoch": 0.08253094910591471,
      "grad_norm": 0.30647218227386475,
      "learning_rate": 0.0009928486058062827,
      "loss": 2.247,
      "step": 180
    },
    {
      "epoch": 0.08298945437872536,
      "grad_norm": 0.27754637598991394,
      "learning_rate": 0.0009927228990299999,
      "loss": 2.2674,
      "step": 181
    },
    {
      "epoch": 0.083447959651536,
      "grad_norm": 0.3044029176235199,
      "learning_rate": 0.0009925961051232468,
      "loss": 2.2634,
      "step": 182
    },
    {
      "epoch": 0.08390646492434663,
      "grad_norm": 0.28203487396240234,
      "learning_rate": 0.000992468224365778,
      "loss": 2.1836,
      "step": 183
    },
    {
      "epoch": 0.08436497019715727,
      "grad_norm": 0.3100438416004181,
      "learning_rate": 0.000992339257039746,
      "loss": 2.2553,
      "step": 184
    },
    {
      "epoch": 0.08482347546996791,
      "grad_norm": 0.27551230788230896,
      "learning_rate": 0.0009922092034297006,
      "loss": 2.1958,
      "step": 185
    },
    {
      "epoch": 0.08528198074277854,
      "grad_norm": 0.26914599537849426,
      "learning_rate": 0.0009920780638225891,
      "loss": 2.1993,
      "step": 186
    },
    {
      "epoch": 0.08574048601558917,
      "grad_norm": 0.29731857776641846,
      "learning_rate": 0.0009919458385077538,
      "loss": 2.1941,
      "step": 187
    },
    {
      "epoch": 0.08619899128839982,
      "grad_norm": 0.28186798095703125,
      "learning_rate": 0.0009918125277769336,
      "loss": 2.1932,
      "step": 188
    },
    {
      "epoch": 0.08665749656121045,
      "grad_norm": 0.2776920199394226,
      "learning_rate": 0.0009916781319242614,
      "loss": 2.2316,
      "step": 189
    },
    {
      "epoch": 0.08711600183402109,
      "grad_norm": 0.26821252703666687,
      "learning_rate": 0.0009915426512462646,
      "loss": 2.2171,
      "step": 190
    },
    {
      "epoch": 0.08757450710683173,
      "grad_norm": 0.28287333250045776,
      "learning_rate": 0.0009914060860418644,
      "loss": 2.2164,
      "step": 191
    },
    {
      "epoch": 0.08803301237964237,
      "grad_norm": 0.2741101086139679,
      "learning_rate": 0.000991268436612374,
      "loss": 2.2648,
      "step": 192
    },
    {
      "epoch": 0.088491517652453,
      "grad_norm": 0.2873133718967438,
      "learning_rate": 0.0009911297032614997,
      "loss": 2.3486,
      "step": 193
    },
    {
      "epoch": 0.08895002292526365,
      "grad_norm": 0.3202584385871887,
      "learning_rate": 0.000990989886295339,
      "loss": 2.1744,
      "step": 194
    },
    {
      "epoch": 0.08940852819807428,
      "grad_norm": 0.29031509160995483,
      "learning_rate": 0.0009908489860223804,
      "loss": 2.262,
      "step": 195
    },
    {
      "epoch": 0.08986703347088491,
      "grad_norm": 0.29200971126556396,
      "learning_rate": 0.000990707002753502,
      "loss": 2.2186,
      "step": 196
    },
    {
      "epoch": 0.09032553874369555,
      "grad_norm": 0.2675837576389313,
      "learning_rate": 0.0009905639368019724,
      "loss": 2.2164,
      "step": 197
    },
    {
      "epoch": 0.0907840440165062,
      "grad_norm": 0.3009452521800995,
      "learning_rate": 0.0009904197884834482,
      "loss": 2.2331,
      "step": 198
    },
    {
      "epoch": 0.09124254928931683,
      "grad_norm": 0.2712272107601166,
      "learning_rate": 0.0009902745581159742,
      "loss": 2.22,
      "step": 199
    },
    {
      "epoch": 0.09170105456212746,
      "grad_norm": 0.2802372872829437,
      "learning_rate": 0.0009901282460199829,
      "loss": 2.2663,
      "step": 200
    },
    {
      "epoch": 0.0921595598349381,
      "grad_norm": 0.27972620725631714,
      "learning_rate": 0.0009899808525182935,
      "loss": 2.2801,
      "step": 201
    },
    {
      "epoch": 0.09261806510774874,
      "grad_norm": 0.2566077709197998,
      "learning_rate": 0.0009898323779361107,
      "loss": 2.1432,
      "step": 202
    },
    {
      "epoch": 0.09307657038055937,
      "grad_norm": 0.2845911681652069,
      "learning_rate": 0.000989682822601025,
      "loss": 2.1841,
      "step": 203
    },
    {
      "epoch": 0.09353507565337002,
      "grad_norm": 0.2847119867801666,
      "learning_rate": 0.0009895321868430113,
      "loss": 2.2173,
      "step": 204
    },
    {
      "epoch": 0.09399358092618065,
      "grad_norm": 0.28393447399139404,
      "learning_rate": 0.0009893804709944281,
      "loss": 2.2069,
      "step": 205
    },
    {
      "epoch": 0.09445208619899129,
      "grad_norm": 0.2812255620956421,
      "learning_rate": 0.0009892276753900174,
      "loss": 2.2182,
      "step": 206
    },
    {
      "epoch": 0.09491059147180192,
      "grad_norm": 0.32794100046157837,
      "learning_rate": 0.0009890738003669028,
      "loss": 2.2537,
      "step": 207
    },
    {
      "epoch": 0.09536909674461257,
      "grad_norm": 0.27341362833976746,
      "learning_rate": 0.0009889188462645904,
      "loss": 2.2562,
      "step": 208
    },
    {
      "epoch": 0.0958276020174232,
      "grad_norm": 0.30244219303131104,
      "learning_rate": 0.0009887628134249667,
      "loss": 2.2801,
      "step": 209
    },
    {
      "epoch": 0.09628610729023383,
      "grad_norm": 0.2909850776195526,
      "learning_rate": 0.0009886057021922983,
      "loss": 2.3029,
      "step": 210
    },
    {
      "epoch": 0.09674461256304448,
      "grad_norm": 0.2840065062046051,
      "learning_rate": 0.0009884475129132311,
      "loss": 2.2221,
      "step": 211
    },
    {
      "epoch": 0.09720311783585511,
      "grad_norm": 0.2807883620262146,
      "learning_rate": 0.0009882882459367897,
      "loss": 2.1279,
      "step": 212
    },
    {
      "epoch": 0.09766162310866575,
      "grad_norm": 0.3041750490665436,
      "learning_rate": 0.0009881279016143766,
      "loss": 2.0585,
      "step": 213
    },
    {
      "epoch": 0.09812012838147639,
      "grad_norm": 0.28147101402282715,
      "learning_rate": 0.0009879664802997707,
      "loss": 2.2139,
      "step": 214
    },
    {
      "epoch": 0.09857863365428703,
      "grad_norm": 0.28452959656715393,
      "learning_rate": 0.000987803982349128,
      "loss": 2.2637,
      "step": 215
    },
    {
      "epoch": 0.09903713892709766,
      "grad_norm": 0.27389010787010193,
      "learning_rate": 0.0009876404081209796,
      "loss": 2.2598,
      "step": 216
    },
    {
      "epoch": 0.09949564419990829,
      "grad_norm": 0.2591685950756073,
      "learning_rate": 0.000987475757976231,
      "loss": 2.1974,
      "step": 217
    },
    {
      "epoch": 0.09995414947271894,
      "grad_norm": 0.2577989399433136,
      "learning_rate": 0.000987310032278162,
      "loss": 2.1759,
      "step": 218
    },
    {
      "epoch": 0.10041265474552957,
      "grad_norm": 0.2675469219684601,
      "learning_rate": 0.0009871432313924254,
      "loss": 2.1756,
      "step": 219
    },
    {
      "epoch": 0.1008711600183402,
      "grad_norm": 0.2864185571670532,
      "learning_rate": 0.000986975355687046,
      "loss": 2.2112,
      "step": 220
    },
    {
      "epoch": 0.10132966529115085,
      "grad_norm": 0.2839012145996094,
      "learning_rate": 0.0009868064055324204,
      "loss": 2.221,
      "step": 221
    },
    {
      "epoch": 0.10178817056396149,
      "grad_norm": 0.2771560549736023,
      "learning_rate": 0.0009866363813013153,
      "loss": 2.2332,
      "step": 222
    },
    {
      "epoch": 0.10224667583677212,
      "grad_norm": 0.260655015707016,
      "learning_rate": 0.0009864652833688676,
      "loss": 2.1928,
      "step": 223
    },
    {
      "epoch": 0.10270518110958277,
      "grad_norm": 0.2673497498035431,
      "learning_rate": 0.0009862931121125836,
      "loss": 2.1672,
      "step": 224
    },
    {
      "epoch": 0.1031636863823934,
      "grad_norm": 0.2705249488353729,
      "learning_rate": 0.000986119867912337,
      "loss": 2.1987,
      "step": 225
    },
    {
      "epoch": 0.10362219165520403,
      "grad_norm": 0.2638174295425415,
      "learning_rate": 0.000985945551150369,
      "loss": 2.2033,
      "step": 226
    },
    {
      "epoch": 0.10408069692801467,
      "grad_norm": 0.260036826133728,
      "learning_rate": 0.0009857701622112876,
      "loss": 2.2107,
      "step": 227
    },
    {
      "epoch": 0.10453920220082531,
      "grad_norm": 0.2825823128223419,
      "learning_rate": 0.000985593701482066,
      "loss": 2.2497,
      "step": 228
    },
    {
      "epoch": 0.10499770747363595,
      "grad_norm": 0.26356783509254456,
      "learning_rate": 0.0009854161693520424,
      "loss": 2.1109,
      "step": 229
    },
    {
      "epoch": 0.10545621274644658,
      "grad_norm": 0.2668766677379608,
      "learning_rate": 0.0009852375662129194,
      "loss": 2.1966,
      "step": 230
    },
    {
      "epoch": 0.10591471801925723,
      "grad_norm": 0.2763180136680603,
      "learning_rate": 0.0009850578924587613,
      "loss": 2.1675,
      "step": 231
    },
    {
      "epoch": 0.10637322329206786,
      "grad_norm": 0.2713163197040558,
      "learning_rate": 0.000984877148485996,
      "loss": 2.2504,
      "step": 232
    },
    {
      "epoch": 0.10683172856487849,
      "grad_norm": 0.300942599773407,
      "learning_rate": 0.000984695334693412,
      "loss": 2.2895,
      "step": 233
    },
    {
      "epoch": 0.10729023383768914,
      "grad_norm": 0.28864598274230957,
      "learning_rate": 0.000984512451482158,
      "loss": 2.2348,
      "step": 234
    },
    {
      "epoch": 0.10774873911049977,
      "grad_norm": 0.2590393126010895,
      "learning_rate": 0.0009843284992557431,
      "loss": 2.1867,
      "step": 235
    },
    {
      "epoch": 0.1082072443833104,
      "grad_norm": 0.26678600907325745,
      "learning_rate": 0.000984143478420034,
      "loss": 2.2365,
      "step": 236
    },
    {
      "epoch": 0.10866574965612105,
      "grad_norm": 0.2753410339355469,
      "learning_rate": 0.0009839573893832563,
      "loss": 2.1442,
      "step": 237
    },
    {
      "epoch": 0.10912425492893169,
      "grad_norm": 0.2811972200870514,
      "learning_rate": 0.000983770232555991,
      "loss": 2.2651,
      "step": 238
    },
    {
      "epoch": 0.10958276020174232,
      "grad_norm": 0.2604229152202606,
      "learning_rate": 0.0009835820083511765,
      "loss": 2.1664,
      "step": 239
    },
    {
      "epoch": 0.11004126547455295,
      "grad_norm": 0.2630242109298706,
      "learning_rate": 0.0009833927171841055,
      "loss": 2.2816,
      "step": 240
    },
    {
      "epoch": 0.1104997707473636,
      "grad_norm": 0.2730475068092346,
      "learning_rate": 0.0009832023594724246,
      "loss": 2.2673,
      "step": 241
    },
    {
      "epoch": 0.11095827602017423,
      "grad_norm": 0.25733426213264465,
      "learning_rate": 0.0009830109356361344,
      "loss": 2.2102,
      "step": 242
    },
    {
      "epoch": 0.11141678129298486,
      "grad_norm": 0.24431732296943665,
      "learning_rate": 0.0009828184460975867,
      "loss": 2.2246,
      "step": 243
    },
    {
      "epoch": 0.11187528656579551,
      "grad_norm": 0.23752157390117645,
      "learning_rate": 0.0009826248912814855,
      "loss": 2.1543,
      "step": 244
    },
    {
      "epoch": 0.11233379183860615,
      "grad_norm": 0.26304981112480164,
      "learning_rate": 0.0009824302716148847,
      "loss": 2.1908,
      "step": 245
    },
    {
      "epoch": 0.11279229711141678,
      "grad_norm": 0.27463531494140625,
      "learning_rate": 0.0009822345875271884,
      "loss": 2.1731,
      "step": 246
    },
    {
      "epoch": 0.11325080238422743,
      "grad_norm": 0.2856203019618988,
      "learning_rate": 0.0009820378394501481,
      "loss": 2.1667,
      "step": 247
    },
    {
      "epoch": 0.11370930765703806,
      "grad_norm": 0.28359538316726685,
      "learning_rate": 0.0009818400278178636,
      "loss": 2.1612,
      "step": 248
    },
    {
      "epoch": 0.11416781292984869,
      "grad_norm": 0.2578238546848297,
      "learning_rate": 0.0009816411530667814,
      "loss": 2.1904,
      "step": 249
    },
    {
      "epoch": 0.11462631820265932,
      "grad_norm": 0.27736279368400574,
      "learning_rate": 0.000981441215635693,
      "loss": 2.2346,
      "step": 250
    },
    {
      "epoch": 0.11508482347546997,
      "grad_norm": 0.2668209671974182,
      "learning_rate": 0.0009812402159657353,
      "loss": 2.2408,
      "step": 251
    },
    {
      "epoch": 0.1155433287482806,
      "grad_norm": 0.2662009000778198,
      "learning_rate": 0.000981038154500388,
      "loss": 2.2384,
      "step": 252
    },
    {
      "epoch": 0.11600183402109124,
      "grad_norm": 0.2680983245372772,
      "learning_rate": 0.0009808350316854746,
      "loss": 2.1905,
      "step": 253
    },
    {
      "epoch": 0.11646033929390188,
      "grad_norm": 0.2629610598087311,
      "learning_rate": 0.0009806308479691594,
      "loss": 2.2658,
      "step": 254
    },
    {
      "epoch": 0.11691884456671252,
      "grad_norm": 0.2751751244068146,
      "learning_rate": 0.0009804256038019482,
      "loss": 2.1762,
      "step": 255
    },
    {
      "epoch": 0.11737734983952315,
      "grad_norm": 0.2628040015697479,
      "learning_rate": 0.0009802192996366857,
      "loss": 2.2191,
      "step": 256
    },
    {
      "epoch": 0.1178358551123338,
      "grad_norm": 0.2667428255081177,
      "learning_rate": 0.0009800119359285563,
      "loss": 2.1598,
      "step": 257
    },
    {
      "epoch": 0.11829436038514443,
      "grad_norm": 0.2692457437515259,
      "learning_rate": 0.0009798035131350813,
      "loss": 2.207,
      "step": 258
    },
    {
      "epoch": 0.11875286565795506,
      "grad_norm": 0.2623954713344574,
      "learning_rate": 0.0009795940317161194,
      "loss": 2.1794,
      "step": 259
    },
    {
      "epoch": 0.1192113709307657,
      "grad_norm": 0.26822301745414734,
      "learning_rate": 0.0009793834921338646,
      "loss": 2.255,
      "step": 260
    },
    {
      "epoch": 0.11966987620357634,
      "grad_norm": 0.2658577859401703,
      "learning_rate": 0.0009791718948528457,
      "loss": 2.2963,
      "step": 261
    },
    {
      "epoch": 0.12012838147638698,
      "grad_norm": 0.2575671374797821,
      "learning_rate": 0.0009789592403399252,
      "loss": 2.1645,
      "step": 262
    },
    {
      "epoch": 0.12058688674919761,
      "grad_norm": 0.2533425986766815,
      "learning_rate": 0.0009787455290642985,
      "loss": 2.1667,
      "step": 263
    },
    {
      "epoch": 0.12104539202200826,
      "grad_norm": 0.2814098000526428,
      "learning_rate": 0.000978530761497492,
      "loss": 2.1934,
      "step": 264
    },
    {
      "epoch": 0.12150389729481889,
      "grad_norm": 0.26615622639656067,
      "learning_rate": 0.0009783149381133633,
      "loss": 2.1563,
      "step": 265
    },
    {
      "epoch": 0.12196240256762952,
      "grad_norm": 0.2619237005710602,
      "learning_rate": 0.0009780980593880992,
      "loss": 2.1465,
      "step": 266
    },
    {
      "epoch": 0.12242090784044017,
      "grad_norm": 0.31059688329696655,
      "learning_rate": 0.0009778801258002153,
      "loss": 2.1835,
      "step": 267
    },
    {
      "epoch": 0.1228794131132508,
      "grad_norm": 0.25823381543159485,
      "learning_rate": 0.000977661137830554,
      "loss": 2.1885,
      "step": 268
    },
    {
      "epoch": 0.12333791838606144,
      "grad_norm": 0.2833395302295685,
      "learning_rate": 0.0009774410959622845,
      "loss": 2.1943,
      "step": 269
    },
    {
      "epoch": 0.12379642365887207,
      "grad_norm": 0.27451157569885254,
      "learning_rate": 0.000977220000680901,
      "loss": 2.0871,
      "step": 270
    },
    {
      "epoch": 0.12425492893168272,
      "grad_norm": 0.2545366585254669,
      "learning_rate": 0.000976997852474223,
      "loss": 2.1037,
      "step": 271
    },
    {
      "epoch": 0.12471343420449335,
      "grad_norm": 0.26117977499961853,
      "learning_rate": 0.0009767746518323914,
      "loss": 2.1177,
      "step": 272
    },
    {
      "epoch": 0.12517193947730398,
      "grad_norm": 0.2645549774169922,
      "learning_rate": 0.0009765503992478704,
      "loss": 2.1248,
      "step": 273
    },
    {
      "epoch": 0.12563044475011462,
      "grad_norm": 0.2689288556575775,
      "learning_rate": 0.0009763250952154449,
      "loss": 2.1587,
      "step": 274
    },
    {
      "epoch": 0.12608895002292525,
      "grad_norm": 0.26478666067123413,
      "learning_rate": 0.0009760987402322195,
      "loss": 2.1654,
      "step": 275
    },
    {
      "epoch": 0.1265474552957359,
      "grad_norm": 0.2723810076713562,
      "learning_rate": 0.0009758713347976178,
      "loss": 2.2387,
      "step": 276
    },
    {
      "epoch": 0.12700596056854654,
      "grad_norm": 0.255545437335968,
      "learning_rate": 0.000975642879413381,
      "loss": 2.1934,
      "step": 277
    },
    {
      "epoch": 0.12746446584135718,
      "grad_norm": 0.2560661733150482,
      "learning_rate": 0.0009754133745835665,
      "loss": 2.1986,
      "step": 278
    },
    {
      "epoch": 0.1279229711141678,
      "grad_norm": 0.24829904735088348,
      "learning_rate": 0.0009751828208145482,
      "loss": 2.1774,
      "step": 279
    },
    {
      "epoch": 0.12838147638697844,
      "grad_norm": 0.2673141658306122,
      "learning_rate": 0.0009749512186150131,
      "loss": 2.2164,
      "step": 280
    },
    {
      "epoch": 0.12883998165978908,
      "grad_norm": 0.26323482394218445,
      "learning_rate": 0.0009747185684959625,
      "loss": 2.2103,
      "step": 281
    },
    {
      "epoch": 0.12929848693259974,
      "grad_norm": 0.2583684027194977,
      "learning_rate": 0.000974484870970709,
      "loss": 2.1815,
      "step": 282
    },
    {
      "epoch": 0.12975699220541037,
      "grad_norm": 0.2638012170791626,
      "learning_rate": 0.0009742501265548767,
      "loss": 2.2139,
      "step": 283
    },
    {
      "epoch": 0.130215497478221,
      "grad_norm": 0.2567639648914337,
      "learning_rate": 0.0009740143357663993,
      "loss": 2.2499,
      "step": 284
    },
    {
      "epoch": 0.13067400275103164,
      "grad_norm": 0.26872798800468445,
      "learning_rate": 0.000973777499125519,
      "loss": 2.1775,
      "step": 285
    },
    {
      "epoch": 0.13113250802384227,
      "grad_norm": 0.2525768280029297,
      "learning_rate": 0.0009735396171547859,
      "loss": 2.1293,
      "step": 286
    },
    {
      "epoch": 0.1315910132966529,
      "grad_norm": 0.25354015827178955,
      "learning_rate": 0.0009733006903790564,
      "loss": 2.2247,
      "step": 287
    },
    {
      "epoch": 0.13204951856946354,
      "grad_norm": 0.2593800127506256,
      "learning_rate": 0.0009730607193254922,
      "loss": 2.2184,
      "step": 288
    },
    {
      "epoch": 0.1325080238422742,
      "grad_norm": 0.24703963100910187,
      "learning_rate": 0.0009728197045235585,
      "loss": 2.209,
      "step": 289
    },
    {
      "epoch": 0.13296652911508483,
      "grad_norm": 0.26162073016166687,
      "learning_rate": 0.0009725776465050242,
      "loss": 2.1628,
      "step": 290
    },
    {
      "epoch": 0.13342503438789546,
      "grad_norm": 0.2693421542644501,
      "learning_rate": 0.0009723345458039594,
      "loss": 2.1735,
      "step": 291
    },
    {
      "epoch": 0.1338835396607061,
      "grad_norm": 0.28077802062034607,
      "learning_rate": 0.000972090402956735,
      "loss": 2.1983,
      "step": 292
    },
    {
      "epoch": 0.13434204493351673,
      "grad_norm": 0.26197972893714905,
      "learning_rate": 0.0009718452185020212,
      "loss": 2.165,
      "step": 293
    },
    {
      "epoch": 0.13480055020632736,
      "grad_norm": 0.26868757605552673,
      "learning_rate": 0.0009715989929807862,
      "loss": 2.1411,
      "step": 294
    },
    {
      "epoch": 0.13525905547913802,
      "grad_norm": 0.2756407856941223,
      "learning_rate": 0.0009713517269362955,
      "loss": 2.2114,
      "step": 295
    },
    {
      "epoch": 0.13571756075194866,
      "grad_norm": 0.25120407342910767,
      "learning_rate": 0.0009711034209141101,
      "loss": 2.1464,
      "step": 296
    },
    {
      "epoch": 0.1361760660247593,
      "grad_norm": 0.2810468375682831,
      "learning_rate": 0.0009708540754620856,
      "loss": 2.2379,
      "step": 297
    },
    {
      "epoch": 0.13663457129756992,
      "grad_norm": 0.2639043927192688,
      "learning_rate": 0.0009706036911303713,
      "loss": 2.2322,
      "step": 298
    },
    {
      "epoch": 0.13709307657038056,
      "grad_norm": 0.2564607560634613,
      "learning_rate": 0.0009703522684714083,
      "loss": 2.1311,
      "step": 299
    },
    {
      "epoch": 0.1375515818431912,
      "grad_norm": 0.24256686866283417,
      "learning_rate": 0.0009700998080399286,
      "loss": 2.1878,
      "step": 300
    },
    {
      "epoch": 0.13801008711600182,
      "grad_norm": 0.27045148611068726,
      "learning_rate": 0.0009698463103929542,
      "loss": 2.2081,
      "step": 301
    },
    {
      "epoch": 0.13846859238881248,
      "grad_norm": 0.24499888718128204,
      "learning_rate": 0.0009695917760897954,
      "loss": 2.1766,
      "step": 302
    },
    {
      "epoch": 0.13892709766162312,
      "grad_norm": 0.24966712296009064,
      "learning_rate": 0.0009693362056920501,
      "loss": 2.1874,
      "step": 303
    },
    {
      "epoch": 0.13938560293443375,
      "grad_norm": 0.23815707862377167,
      "learning_rate": 0.0009690795997636015,
      "loss": 2.1777,
      "step": 304
    },
    {
      "epoch": 0.13984410820724438,
      "grad_norm": 0.24883851408958435,
      "learning_rate": 0.0009688219588706179,
      "loss": 2.1346,
      "step": 305
    },
    {
      "epoch": 0.14030261348005502,
      "grad_norm": 0.2473696917295456,
      "learning_rate": 0.0009685632835815518,
      "loss": 2.1619,
      "step": 306
    },
    {
      "epoch": 0.14076111875286565,
      "grad_norm": 0.25739142298698425,
      "learning_rate": 0.0009683035744671367,
      "loss": 2.1696,
      "step": 307
    },
    {
      "epoch": 0.14121962402567628,
      "grad_norm": 0.251176118850708,
      "learning_rate": 0.0009680428321003883,
      "loss": 2.1694,
      "step": 308
    },
    {
      "epoch": 0.14167812929848694,
      "grad_norm": 0.25369200110435486,
      "learning_rate": 0.000967781057056601,
      "loss": 2.1791,
      "step": 309
    },
    {
      "epoch": 0.14213663457129758,
      "grad_norm": 0.2517925798892975,
      "learning_rate": 0.0009675182499133485,
      "loss": 2.2125,
      "step": 310
    },
    {
      "epoch": 0.1425951398441082,
      "grad_norm": 0.2633146345615387,
      "learning_rate": 0.0009672544112504813,
      "loss": 2.1691,
      "step": 311
    },
    {
      "epoch": 0.14305364511691884,
      "grad_norm": 0.24655553698539734,
      "learning_rate": 0.0009669895416501257,
      "loss": 2.125,
      "step": 312
    },
    {
      "epoch": 0.14351215038972948,
      "grad_norm": 0.26560506224632263,
      "learning_rate": 0.0009667236416966833,
      "loss": 2.218,
      "step": 313
    },
    {
      "epoch": 0.1439706556625401,
      "grad_norm": 0.26598498225212097,
      "learning_rate": 0.0009664567119768281,
      "loss": 2.1744,
      "step": 314
    },
    {
      "epoch": 0.14442916093535077,
      "grad_norm": 0.2449737787246704,
      "learning_rate": 0.0009661887530795067,
      "loss": 2.1153,
      "step": 315
    },
    {
      "epoch": 0.1448876662081614,
      "grad_norm": 0.2651059925556183,
      "learning_rate": 0.0009659197655959365,
      "loss": 2.1717,
      "step": 316
    },
    {
      "epoch": 0.14534617148097204,
      "grad_norm": 0.2613866627216339,
      "learning_rate": 0.000965649750119604,
      "loss": 2.1991,
      "step": 317
    },
    {
      "epoch": 0.14580467675378267,
      "grad_norm": 0.26489102840423584,
      "learning_rate": 0.0009653787072462643,
      "loss": 2.1615,
      "step": 318
    },
    {
      "epoch": 0.1462631820265933,
      "grad_norm": 0.23147818446159363,
      "learning_rate": 0.0009651066375739388,
      "loss": 2.2432,
      "step": 319
    },
    {
      "epoch": 0.14672168729940394,
      "grad_norm": 0.24258069694042206,
      "learning_rate": 0.000964833541702915,
      "loss": 2.1322,
      "step": 320
    },
    {
      "epoch": 0.14718019257221457,
      "grad_norm": 0.23859402537345886,
      "learning_rate": 0.0009645594202357438,
      "loss": 2.1911,
      "step": 321
    },
    {
      "epoch": 0.14763869784502523,
      "grad_norm": 0.244167298078537,
      "learning_rate": 0.0009642842737772397,
      "loss": 2.1642,
      "step": 322
    },
    {
      "epoch": 0.14809720311783586,
      "grad_norm": 0.24577149748802185,
      "learning_rate": 0.0009640081029344782,
      "loss": 2.1645,
      "step": 323
    },
    {
      "epoch": 0.1485557083906465,
      "grad_norm": 0.2628529369831085,
      "learning_rate": 0.0009637309083167956,
      "loss": 2.229,
      "step": 324
    },
    {
      "epoch": 0.14901421366345713,
      "grad_norm": 0.24074026942253113,
      "learning_rate": 0.0009634526905357859,
      "loss": 2.1841,
      "step": 325
    },
    {
      "epoch": 0.14947271893626776,
      "grad_norm": 0.24728497862815857,
      "learning_rate": 0.000963173450205302,
      "loss": 2.2116,
      "step": 326
    },
    {
      "epoch": 0.1499312242090784,
      "grad_norm": 0.23879386484622955,
      "learning_rate": 0.0009628931879414517,
      "loss": 2.1098,
      "step": 327
    },
    {
      "epoch": 0.15038972948188903,
      "grad_norm": 0.27072983980178833,
      "learning_rate": 0.0009626119043625983,
      "loss": 2.2262,
      "step": 328
    },
    {
      "epoch": 0.1508482347546997,
      "grad_norm": 0.2474377602338791,
      "learning_rate": 0.0009623296000893582,
      "loss": 2.1148,
      "step": 329
    },
    {
      "epoch": 0.15130674002751032,
      "grad_norm": 0.25553756952285767,
      "learning_rate": 0.0009620462757446,
      "loss": 2.1431,
      "step": 330
    },
    {
      "epoch": 0.15176524530032096,
      "grad_norm": 0.25891098380088806,
      "learning_rate": 0.0009617619319534428,
      "loss": 2.2141,
      "step": 331
    },
    {
      "epoch": 0.1522237505731316,
      "grad_norm": 0.2547933757305145,
      "learning_rate": 0.000961476569343255,
      "loss": 2.1754,
      "step": 332
    },
    {
      "epoch": 0.15268225584594222,
      "grad_norm": 0.23339241743087769,
      "learning_rate": 0.0009611901885436529,
      "loss": 2.1727,
      "step": 333
    },
    {
      "epoch": 0.15314076111875286,
      "grad_norm": 0.23726585507392883,
      "learning_rate": 0.0009609027901864996,
      "loss": 2.1151,
      "step": 334
    },
    {
      "epoch": 0.15359926639156352,
      "grad_norm": 0.2518211603164673,
      "learning_rate": 0.0009606143749059029,
      "loss": 2.169,
      "step": 335
    },
    {
      "epoch": 0.15405777166437415,
      "grad_norm": 0.2564687430858612,
      "learning_rate": 0.0009603249433382144,
      "loss": 2.1894,
      "step": 336
    },
    {
      "epoch": 0.15451627693718478,
      "grad_norm": 0.27494797110557556,
      "learning_rate": 0.0009600344961220282,
      "loss": 2.1938,
      "step": 337
    },
    {
      "epoch": 0.15497478220999542,
      "grad_norm": 0.24538889527320862,
      "learning_rate": 0.0009597430338981791,
      "loss": 2.177,
      "step": 338
    },
    {
      "epoch": 0.15543328748280605,
      "grad_norm": 0.23032133281230927,
      "learning_rate": 0.0009594505573097414,
      "loss": 2.1775,
      "step": 339
    },
    {
      "epoch": 0.15589179275561668,
      "grad_norm": 0.24556130170822144,
      "learning_rate": 0.0009591570670020277,
      "loss": 2.2047,
      "step": 340
    },
    {
      "epoch": 0.15635029802842731,
      "grad_norm": 0.2423062026500702,
      "learning_rate": 0.0009588625636225871,
      "loss": 2.1353,
      "step": 341
    },
    {
      "epoch": 0.15680880330123798,
      "grad_norm": 0.24103741347789764,
      "learning_rate": 0.0009585670478212036,
      "loss": 2.0202,
      "step": 342
    },
    {
      "epoch": 0.1572673085740486,
      "grad_norm": 0.24493791162967682,
      "learning_rate": 0.0009582705202498956,
      "loss": 2.1634,
      "step": 343
    },
    {
      "epoch": 0.15772581384685924,
      "grad_norm": 0.24880501627922058,
      "learning_rate": 0.0009579729815629133,
      "loss": 2.2067,
      "step": 344
    },
    {
      "epoch": 0.15818431911966988,
      "grad_norm": 0.24915951490402222,
      "learning_rate": 0.0009576744324167379,
      "loss": 2.2664,
      "step": 345
    },
    {
      "epoch": 0.1586428243924805,
      "grad_norm": 0.23547308146953583,
      "learning_rate": 0.0009573748734700804,
      "loss": 2.0969,
      "step": 346
    },
    {
      "epoch": 0.15910132966529114,
      "grad_norm": 0.2410992532968521,
      "learning_rate": 0.0009570743053838796,
      "loss": 2.2274,
      "step": 347
    },
    {
      "epoch": 0.15955983493810177,
      "grad_norm": 0.24414683878421783,
      "learning_rate": 0.0009567727288213005,
      "loss": 2.1281,
      "step": 348
    },
    {
      "epoch": 0.16001834021091244,
      "grad_norm": 0.23081733286380768,
      "learning_rate": 0.0009564701444477337,
      "loss": 2.1952,
      "step": 349
    },
    {
      "epoch": 0.16047684548372307,
      "grad_norm": 0.2316802740097046,
      "learning_rate": 0.000956166552930793,
      "loss": 2.1237,
      "step": 350
    },
    {
      "epoch": 0.1609353507565337,
      "grad_norm": 0.23908056318759918,
      "learning_rate": 0.0009558619549403147,
      "loss": 2.2821,
      "step": 351
    },
    {
      "epoch": 0.16139385602934433,
      "grad_norm": 0.2308858036994934,
      "learning_rate": 0.0009555563511483555,
      "loss": 2.1321,
      "step": 352
    },
    {
      "epoch": 0.16185236130215497,
      "grad_norm": 0.23893655836582184,
      "learning_rate": 0.0009552497422291912,
      "loss": 2.1284,
      "step": 353
    },
    {
      "epoch": 0.1623108665749656,
      "grad_norm": 0.2301374226808548,
      "learning_rate": 0.0009549421288593157,
      "loss": 2.2148,
      "step": 354
    },
    {
      "epoch": 0.16276937184777626,
      "grad_norm": 0.22996391355991364,
      "learning_rate": 0.0009546335117174385,
      "loss": 2.2119,
      "step": 355
    },
    {
      "epoch": 0.1632278771205869,
      "grad_norm": 0.24970418214797974,
      "learning_rate": 0.0009543238914844843,
      "loss": 2.2142,
      "step": 356
    },
    {
      "epoch": 0.16368638239339753,
      "grad_norm": 0.2592678666114807,
      "learning_rate": 0.0009540132688435907,
      "loss": 2.1741,
      "step": 357
    },
    {
      "epoch": 0.16414488766620816,
      "grad_norm": 0.26019734144210815,
      "learning_rate": 0.0009537016444801074,
      "loss": 2.0638,
      "step": 358
    },
    {
      "epoch": 0.1646033929390188,
      "grad_norm": 0.25875401496887207,
      "learning_rate": 0.0009533890190815935,
      "loss": 2.2819,
      "step": 359
    },
    {
      "epoch": 0.16506189821182943,
      "grad_norm": 0.24668289721012115,
      "learning_rate": 0.0009530753933378173,
      "loss": 2.1445,
      "step": 360
    },
    {
      "epoch": 0.16552040348464006,
      "grad_norm": 0.22949308156967163,
      "learning_rate": 0.0009527607679407545,
      "loss": 2.1611,
      "step": 361
    },
    {
      "epoch": 0.16597890875745072,
      "grad_norm": 0.22246147692203522,
      "learning_rate": 0.0009524451435845857,
      "loss": 2.1807,
      "step": 362
    },
    {
      "epoch": 0.16643741403026135,
      "grad_norm": 0.24279576539993286,
      "learning_rate": 0.0009521285209656963,
      "loss": 2.1175,
      "step": 363
    },
    {
      "epoch": 0.166895919303072,
      "grad_norm": 0.2524352967739105,
      "learning_rate": 0.0009518109007826734,
      "loss": 2.0727,
      "step": 364
    },
    {
      "epoch": 0.16735442457588262,
      "grad_norm": 0.2450723797082901,
      "learning_rate": 0.0009514922837363059,
      "loss": 2.2049,
      "step": 365
    },
    {
      "epoch": 0.16781292984869325,
      "grad_norm": 0.22783881425857544,
      "learning_rate": 0.0009511726705295817,
      "loss": 2.1853,
      "step": 366
    },
    {
      "epoch": 0.1682714351215039,
      "grad_norm": 0.2661840617656708,
      "learning_rate": 0.000950852061867687,
      "loss": 2.1777,
      "step": 367
    },
    {
      "epoch": 0.16872994039431455,
      "grad_norm": 0.2511211633682251,
      "learning_rate": 0.0009505304584580038,
      "loss": 2.1359,
      "step": 368
    },
    {
      "epoch": 0.16918844566712518,
      "grad_norm": 0.2452414482831955,
      "learning_rate": 0.0009502078610101092,
      "loss": 2.2675,
      "step": 369
    },
    {
      "epoch": 0.16964695093993581,
      "grad_norm": 0.2242978811264038,
      "learning_rate": 0.0009498842702357736,
      "loss": 2.1883,
      "step": 370
    },
    {
      "epoch": 0.17010545621274645,
      "grad_norm": 0.24296517670154572,
      "learning_rate": 0.0009495596868489587,
      "loss": 2.1704,
      "step": 371
    },
    {
      "epoch": 0.17056396148555708,
      "grad_norm": 0.22920766472816467,
      "learning_rate": 0.0009492341115658167,
      "loss": 2.1962,
      "step": 372
    },
    {
      "epoch": 0.1710224667583677,
      "grad_norm": 0.23469820618629456,
      "learning_rate": 0.0009489075451046879,
      "loss": 2.1537,
      "step": 373
    },
    {
      "epoch": 0.17148097203117835,
      "grad_norm": 0.23555652797222137,
      "learning_rate": 0.0009485799881861,
      "loss": 2.1081,
      "step": 374
    },
    {
      "epoch": 0.171939477303989,
      "grad_norm": 0.23698078095912933,
      "learning_rate": 0.0009482514415327654,
      "loss": 2.1992,
      "step": 375
    },
    {
      "epoch": 0.17239798257679964,
      "grad_norm": 0.24469910562038422,
      "learning_rate": 0.000947921905869581,
      "loss": 2.1632,
      "step": 376
    },
    {
      "epoch": 0.17285648784961027,
      "grad_norm": 0.25777336955070496,
      "learning_rate": 0.0009475913819236248,
      "loss": 2.2234,
      "step": 377
    },
    {
      "epoch": 0.1733149931224209,
      "grad_norm": 0.25233694911003113,
      "learning_rate": 0.0009472598704241561,
      "loss": 2.1678,
      "step": 378
    },
    {
      "epoch": 0.17377349839523154,
      "grad_norm": 0.24837735295295715,
      "learning_rate": 0.0009469273721026131,
      "loss": 2.1037,
      "step": 379
    },
    {
      "epoch": 0.17423200366804217,
      "grad_norm": 0.24020598828792572,
      "learning_rate": 0.0009465938876926111,
      "loss": 2.0847,
      "step": 380
    },
    {
      "epoch": 0.1746905089408528,
      "grad_norm": 0.24777191877365112,
      "learning_rate": 0.0009462594179299406,
      "loss": 2.2703,
      "step": 381
    },
    {
      "epoch": 0.17514901421366347,
      "grad_norm": 0.2401338368654251,
      "learning_rate": 0.0009459239635525672,
      "loss": 2.1676,
      "step": 382
    },
    {
      "epoch": 0.1756075194864741,
      "grad_norm": 0.24869659543037415,
      "learning_rate": 0.0009455875253006281,
      "loss": 2.1894,
      "step": 383
    },
    {
      "epoch": 0.17606602475928473,
      "grad_norm": 0.24969342350959778,
      "learning_rate": 0.0009452501039164315,
      "loss": 2.2423,
      "step": 384
    },
    {
      "epoch": 0.17652453003209537,
      "grad_norm": 0.24919766187667847,
      "learning_rate": 0.0009449117001444549,
      "loss": 2.1148,
      "step": 385
    },
    {
      "epoch": 0.176983035304906,
      "grad_norm": 0.24014052748680115,
      "learning_rate": 0.0009445723147313433,
      "loss": 2.2137,
      "step": 386
    },
    {
      "epoch": 0.17744154057771663,
      "grad_norm": 0.25087496638298035,
      "learning_rate": 0.0009442319484259074,
      "loss": 2.1282,
      "step": 387
    },
    {
      "epoch": 0.1779000458505273,
      "grad_norm": 0.25125542283058167,
      "learning_rate": 0.0009438906019791222,
      "loss": 2.2208,
      "step": 388
    },
    {
      "epoch": 0.17835855112333793,
      "grad_norm": 0.240860715508461,
      "learning_rate": 0.0009435482761441251,
      "loss": 2.2281,
      "step": 389
    },
    {
      "epoch": 0.17881705639614856,
      "grad_norm": 0.2505457401275635,
      "learning_rate": 0.000943204971676215,
      "loss": 2.2199,
      "step": 390
    },
    {
      "epoch": 0.1792755616689592,
      "grad_norm": 0.25510266423225403,
      "learning_rate": 0.0009428606893328493,
      "loss": 2.1469,
      "step": 391
    },
    {
      "epoch": 0.17973406694176983,
      "grad_norm": 0.23214587569236755,
      "learning_rate": 0.0009425154298736432,
      "loss": 2.1878,
      "step": 392
    },
    {
      "epoch": 0.18019257221458046,
      "grad_norm": 0.22687500715255737,
      "learning_rate": 0.0009421691940603678,
      "loss": 2.1092,
      "step": 393
    },
    {
      "epoch": 0.1806510774873911,
      "grad_norm": 0.2500844895839691,
      "learning_rate": 0.0009418219826569488,
      "loss": 2.2665,
      "step": 394
    },
    {
      "epoch": 0.18110958276020175,
      "grad_norm": 0.24146921932697296,
      "learning_rate": 0.0009414737964294635,
      "loss": 2.0804,
      "step": 395
    },
    {
      "epoch": 0.1815680880330124,
      "grad_norm": 0.24542559683322906,
      "learning_rate": 0.000941124636146141,
      "loss": 2.1867,
      "step": 396
    },
    {
      "epoch": 0.18202659330582302,
      "grad_norm": 0.2595759928226471,
      "learning_rate": 0.0009407745025773589,
      "loss": 2.1253,
      "step": 397
    },
    {
      "epoch": 0.18248509857863365,
      "grad_norm": 0.2373654544353485,
      "learning_rate": 0.0009404233964956423,
      "loss": 2.0456,
      "step": 398
    },
    {
      "epoch": 0.1829436038514443,
      "grad_norm": 0.24088534712791443,
      "learning_rate": 0.0009400713186756625,
      "loss": 2.1647,
      "step": 399
    },
    {
      "epoch": 0.18340210912425492,
      "grad_norm": 0.24192871153354645,
      "learning_rate": 0.0009397182698942342,
      "loss": 2.1178,
      "step": 400
    },
    {
      "epoch": 0.18386061439706555,
      "grad_norm": 0.23445655405521393,
      "learning_rate": 0.0009393642509303149,
      "loss": 2.1694,
      "step": 401
    },
    {
      "epoch": 0.1843191196698762,
      "grad_norm": 0.2317449301481247,
      "learning_rate": 0.0009390092625650023,
      "loss": 2.1357,
      "step": 402
    },
    {
      "epoch": 0.18477762494268685,
      "grad_norm": 0.2417811155319214,
      "learning_rate": 0.0009386533055815332,
      "loss": 2.1377,
      "step": 403
    },
    {
      "epoch": 0.18523613021549748,
      "grad_norm": 0.226463183760643,
      "learning_rate": 0.0009382963807652813,
      "loss": 2.1075,
      "step": 404
    },
    {
      "epoch": 0.1856946354883081,
      "grad_norm": 0.22167465090751648,
      "learning_rate": 0.000937938488903756,
      "loss": 2.1689,
      "step": 405
    },
    {
      "epoch": 0.18615314076111875,
      "grad_norm": 0.23535846173763275,
      "learning_rate": 0.0009375796307866003,
      "loss": 2.1621,
      "step": 406
    },
    {
      "epoch": 0.18661164603392938,
      "grad_norm": 0.21948988735675812,
      "learning_rate": 0.0009372198072055888,
      "loss": 2.1752,
      "step": 407
    },
    {
      "epoch": 0.18707015130674004,
      "grad_norm": 0.2306925356388092,
      "learning_rate": 0.0009368590189546268,
      "loss": 2.1191,
      "step": 408
    },
    {
      "epoch": 0.18752865657955067,
      "grad_norm": 0.23263175785541534,
      "learning_rate": 0.0009364972668297474,
      "loss": 2.1564,
      "step": 409
    },
    {
      "epoch": 0.1879871618523613,
      "grad_norm": 0.24431577324867249,
      "learning_rate": 0.0009361345516291111,
      "loss": 2.135,
      "step": 410
    },
    {
      "epoch": 0.18844566712517194,
      "grad_norm": 0.24164284765720367,
      "learning_rate": 0.0009357708741530024,
      "loss": 2.1717,
      "step": 411
    },
    {
      "epoch": 0.18890417239798257,
      "grad_norm": 0.2662006914615631,
      "learning_rate": 0.00093540623520383,
      "loss": 2.1309,
      "step": 412
    },
    {
      "epoch": 0.1893626776707932,
      "grad_norm": 0.24077622592449188,
      "learning_rate": 0.000935040635586123,
      "loss": 2.1517,
      "step": 413
    },
    {
      "epoch": 0.18982118294360384,
      "grad_norm": 0.24029862880706787,
      "learning_rate": 0.0009346740761065305,
      "loss": 2.1293,
      "step": 414
    },
    {
      "epoch": 0.1902796882164145,
      "grad_norm": 0.24488435685634613,
      "learning_rate": 0.0009343065575738197,
      "loss": 2.1134,
      "step": 415
    },
    {
      "epoch": 0.19073819348922513,
      "grad_norm": 0.22803150117397308,
      "learning_rate": 0.0009339380807988733,
      "loss": 2.1487,
      "step": 416
    },
    {
      "epoch": 0.19119669876203577,
      "grad_norm": 0.2276964783668518,
      "learning_rate": 0.0009335686465946887,
      "loss": 2.1523,
      "step": 417
    },
    {
      "epoch": 0.1916552040348464,
      "grad_norm": 0.252250075340271,
      "learning_rate": 0.0009331982557763754,
      "loss": 2.1931,
      "step": 418
    },
    {
      "epoch": 0.19211370930765703,
      "grad_norm": 0.23511146008968353,
      "learning_rate": 0.0009328269091611537,
      "loss": 2.0303,
      "step": 419
    },
    {
      "epoch": 0.19257221458046767,
      "grad_norm": 0.23385009169578552,
      "learning_rate": 0.0009324546075683524,
      "loss": 2.2053,
      "step": 420
    },
    {
      "epoch": 0.19303071985327833,
      "grad_norm": 0.23939022421836853,
      "learning_rate": 0.0009320813518194083,
      "loss": 2.1458,
      "step": 421
    },
    {
      "epoch": 0.19348922512608896,
      "grad_norm": 0.23134151101112366,
      "learning_rate": 0.0009317071427378624,
      "loss": 2.0712,
      "step": 422
    },
    {
      "epoch": 0.1939477303988996,
      "grad_norm": 0.23936788737773895,
      "learning_rate": 0.0009313319811493594,
      "loss": 2.1149,
      "step": 423
    },
    {
      "epoch": 0.19440623567171023,
      "grad_norm": 0.2574325203895569,
      "learning_rate": 0.000930955867881646,
      "loss": 2.2531,
      "step": 424
    },
    {
      "epoch": 0.19486474094452086,
      "grad_norm": 0.24178549647331238,
      "learning_rate": 0.0009305788037645681,
      "loss": 2.1294,
      "step": 425
    },
    {
      "epoch": 0.1953232462173315,
      "grad_norm": 0.23926787078380585,
      "learning_rate": 0.0009302007896300697,
      "loss": 2.1827,
      "step": 426
    },
    {
      "epoch": 0.19578175149014213,
      "grad_norm": 0.23626065254211426,
      "learning_rate": 0.0009298218263121911,
      "loss": 2.1604,
      "step": 427
    },
    {
      "epoch": 0.19624025676295279,
      "grad_norm": 0.23590530455112457,
      "learning_rate": 0.0009294419146470668,
      "loss": 2.1806,
      "step": 428
    },
    {
      "epoch": 0.19669876203576342,
      "grad_norm": 0.22567199170589447,
      "learning_rate": 0.0009290610554729234,
      "loss": 2.1528,
      "step": 429
    },
    {
      "epoch": 0.19715726730857405,
      "grad_norm": 0.23253653943538666,
      "learning_rate": 0.0009286792496300784,
      "loss": 2.174,
      "step": 430
    },
    {
      "epoch": 0.19761577258138469,
      "grad_norm": 0.21436229348182678,
      "learning_rate": 0.0009282964979609379,
      "loss": 2.1591,
      "step": 431
    },
    {
      "epoch": 0.19807427785419532,
      "grad_norm": 0.23382188379764557,
      "learning_rate": 0.0009279128013099947,
      "loss": 2.1556,
      "step": 432
    },
    {
      "epoch": 0.19853278312700595,
      "grad_norm": 0.2376870959997177,
      "learning_rate": 0.0009275281605238268,
      "loss": 2.1917,
      "step": 433
    },
    {
      "epoch": 0.19899128839981658,
      "grad_norm": 0.23477254807949066,
      "learning_rate": 0.0009271425764510953,
      "loss": 2.1935,
      "step": 434
    },
    {
      "epoch": 0.19944979367262725,
      "grad_norm": 0.22242377698421478,
      "learning_rate": 0.0009267560499425423,
      "loss": 2.082,
      "step": 435
    },
    {
      "epoch": 0.19990829894543788,
      "grad_norm": 0.2337796837091446,
      "learning_rate": 0.0009263685818509895,
      "loss": 2.0812,
      "step": 436
    },
    {
      "epoch": 0.2003668042182485,
      "grad_norm": 0.23238253593444824,
      "learning_rate": 0.000925980173031336,
      "loss": 2.17,
      "step": 437
    },
    {
      "epoch": 0.20082530949105915,
      "grad_norm": 0.23153512179851532,
      "learning_rate": 0.0009255908243405567,
      "loss": 2.1048,
      "step": 438
    },
    {
      "epoch": 0.20128381476386978,
      "grad_norm": 0.24234378337860107,
      "learning_rate": 0.0009252005366376996,
      "loss": 2.2108,
      "step": 439
    },
    {
      "epoch": 0.2017423200366804,
      "grad_norm": 0.23216930031776428,
      "learning_rate": 0.0009248093107838852,
      "loss": 2.155,
      "step": 440
    },
    {
      "epoch": 0.20220082530949107,
      "grad_norm": 0.2328544408082962,
      "learning_rate": 0.0009244171476423036,
      "loss": 2.1477,
      "step": 441
    },
    {
      "epoch": 0.2026593305823017,
      "grad_norm": 0.22742120921611786,
      "learning_rate": 0.0009240240480782129,
      "loss": 2.0584,
      "step": 442
    },
    {
      "epoch": 0.20311783585511234,
      "grad_norm": 0.21954527497291565,
      "learning_rate": 0.0009236300129589376,
      "loss": 2.2378,
      "step": 443
    },
    {
      "epoch": 0.20357634112792297,
      "grad_norm": 0.23131334781646729,
      "learning_rate": 0.0009232350431538657,
      "loss": 2.2118,
      "step": 444
    },
    {
      "epoch": 0.2040348464007336,
      "grad_norm": 0.2409854531288147,
      "learning_rate": 0.0009228391395344482,
      "loss": 2.1398,
      "step": 445
    },
    {
      "epoch": 0.20449335167354424,
      "grad_norm": 0.2221985161304474,
      "learning_rate": 0.000922442302974196,
      "loss": 2.1921,
      "step": 446
    },
    {
      "epoch": 0.20495185694635487,
      "grad_norm": 0.33378636837005615,
      "learning_rate": 0.0009220445343486785,
      "loss": 2.1761,
      "step": 447
    },
    {
      "epoch": 0.20541036221916553,
      "grad_norm": 0.22926853597164154,
      "learning_rate": 0.0009216458345355217,
      "loss": 2.1401,
      "step": 448
    },
    {
      "epoch": 0.20586886749197617,
      "grad_norm": 0.22897504270076752,
      "learning_rate": 0.0009212462044144061,
      "loss": 2.2143,
      "step": 449
    },
    {
      "epoch": 0.2063273727647868,
      "grad_norm": 0.2363223284482956,
      "learning_rate": 0.0009208456448670648,
      "loss": 2.146,
      "step": 450
    },
    {
      "epoch": 0.20678587803759743,
      "grad_norm": 0.22411952912807465,
      "learning_rate": 0.0009204441567772816,
      "loss": 2.2284,
      "step": 451
    },
    {
      "epoch": 0.20724438331040806,
      "grad_norm": 0.21718861162662506,
      "learning_rate": 0.0009200417410308888,
      "loss": 2.0972,
      "step": 452
    },
    {
      "epoch": 0.2077028885832187,
      "grad_norm": 0.2213357388973236,
      "learning_rate": 0.0009196383985157656,
      "loss": 2.1481,
      "step": 453
    },
    {
      "epoch": 0.20816139385602933,
      "grad_norm": 0.21943038702011108,
      "learning_rate": 0.000919234130121836,
      "loss": 2.1292,
      "step": 454
    },
    {
      "epoch": 0.20861989912884,
      "grad_norm": 0.24687804281711578,
      "learning_rate": 0.0009188289367410672,
      "loss": 2.1257,
      "step": 455
    },
    {
      "epoch": 0.20907840440165062,
      "grad_norm": 0.22469475865364075,
      "learning_rate": 0.0009184228192674666,
      "loss": 2.0781,
      "step": 456
    },
    {
      "epoch": 0.20953690967446126,
      "grad_norm": 0.21859581768512726,
      "learning_rate": 0.0009180157785970808,
      "loss": 2.1971,
      "step": 457
    },
    {
      "epoch": 0.2099954149472719,
      "grad_norm": 0.21539609134197235,
      "learning_rate": 0.0009176078156279932,
      "loss": 2.0934,
      "step": 458
    },
    {
      "epoch": 0.21045392022008252,
      "grad_norm": 0.24682052433490753,
      "learning_rate": 0.0009171989312603226,
      "loss": 2.11,
      "step": 459
    },
    {
      "epoch": 0.21091242549289316,
      "grad_norm": 0.22692924737930298,
      "learning_rate": 0.0009167891263962202,
      "loss": 2.148,
      "step": 460
    },
    {
      "epoch": 0.21137093076570382,
      "grad_norm": 0.22987549006938934,
      "learning_rate": 0.0009163784019398685,
      "loss": 2.1468,
      "step": 461
    },
    {
      "epoch": 0.21182943603851445,
      "grad_norm": 0.26675957441329956,
      "learning_rate": 0.0009159667587974785,
      "loss": 2.2705,
      "step": 462
    },
    {
      "epoch": 0.21228794131132508,
      "grad_norm": 0.24283333122730255,
      "learning_rate": 0.0009155541978772887,
      "loss": 2.126,
      "step": 463
    },
    {
      "epoch": 0.21274644658413572,
      "grad_norm": 0.23724466562271118,
      "learning_rate": 0.0009151407200895625,
      "loss": 2.0316,
      "step": 464
    },
    {
      "epoch": 0.21320495185694635,
      "grad_norm": 0.24166786670684814,
      "learning_rate": 0.000914726326346586,
      "loss": 2.0816,
      "step": 465
    },
    {
      "epoch": 0.21366345712975698,
      "grad_norm": 0.2385045737028122,
      "learning_rate": 0.0009143110175626661,
      "loss": 2.2054,
      "step": 466
    },
    {
      "epoch": 0.21412196240256762,
      "grad_norm": 0.23160219192504883,
      "learning_rate": 0.0009138947946541291,
      "loss": 2.2648,
      "step": 467
    },
    {
      "epoch": 0.21458046767537828,
      "grad_norm": 0.2344360649585724,
      "learning_rate": 0.0009134776585393181,
      "loss": 2.114,
      "step": 468
    },
    {
      "epoch": 0.2150389729481889,
      "grad_norm": 0.2413543313741684,
      "learning_rate": 0.0009130596101385906,
      "loss": 2.2026,
      "step": 469
    },
    {
      "epoch": 0.21549747822099954,
      "grad_norm": 0.22701117396354675,
      "learning_rate": 0.0009126406503743174,
      "loss": 2.1773,
      "step": 470
    },
    {
      "epoch": 0.21595598349381018,
      "grad_norm": 0.2317270189523697,
      "learning_rate": 0.0009122207801708802,
      "loss": 2.0533,
      "step": 471
    },
    {
      "epoch": 0.2164144887666208,
      "grad_norm": 0.23005087673664093,
      "learning_rate": 0.0009118000004546689,
      "loss": 2.1755,
      "step": 472
    },
    {
      "epoch": 0.21687299403943144,
      "grad_norm": 0.22558781504631042,
      "learning_rate": 0.0009113783121540807,
      "loss": 2.097,
      "step": 473
    },
    {
      "epoch": 0.2173314993122421,
      "grad_norm": 0.22842420637607574,
      "learning_rate": 0.0009109557161995172,
      "loss": 2.1676,
      "step": 474
    },
    {
      "epoch": 0.21779000458505274,
      "grad_norm": 0.22644348442554474,
      "learning_rate": 0.0009105322135233828,
      "loss": 2.1819,
      "step": 475
    },
    {
      "epoch": 0.21824850985786337,
      "grad_norm": 0.24043895304203033,
      "learning_rate": 0.0009101078050600821,
      "loss": 2.1295,
      "step": 476
    },
    {
      "epoch": 0.218707015130674,
      "grad_norm": 0.24344541132450104,
      "learning_rate": 0.0009096824917460186,
      "loss": 2.1669,
      "step": 477
    },
    {
      "epoch": 0.21916552040348464,
      "grad_norm": 0.24093249440193176,
      "learning_rate": 0.0009092562745195921,
      "loss": 2.1516,
      "step": 478
    },
    {
      "epoch": 0.21962402567629527,
      "grad_norm": 0.23973432183265686,
      "learning_rate": 0.0009088291543211967,
      "loss": 2.1591,
      "step": 479
    },
    {
      "epoch": 0.2200825309491059,
      "grad_norm": 0.23617641627788544,
      "learning_rate": 0.0009084011320932188,
      "loss": 2.1465,
      "step": 480
    },
    {
      "epoch": 0.22054103622191656,
      "grad_norm": 0.22675283253192902,
      "learning_rate": 0.0009079722087800352,
      "loss": 2.0926,
      "step": 481
    },
    {
      "epoch": 0.2209995414947272,
      "grad_norm": 0.233818918466568,
      "learning_rate": 0.0009075423853280106,
      "loss": 2.181,
      "step": 482
    },
    {
      "epoch": 0.22145804676753783,
      "grad_norm": 0.22531652450561523,
      "learning_rate": 0.0009071116626854958,
      "loss": 2.092,
      "step": 483
    },
    {
      "epoch": 0.22191655204034846,
      "grad_norm": 0.23466913402080536,
      "learning_rate": 0.0009066800418028256,
      "loss": 2.1707,
      "step": 484
    },
    {
      "epoch": 0.2223750573131591,
      "grad_norm": 0.21765723824501038,
      "learning_rate": 0.0009062475236323168,
      "loss": 2.1545,
      "step": 485
    },
    {
      "epoch": 0.22283356258596973,
      "grad_norm": 0.22925378382205963,
      "learning_rate": 0.0009058141091282656,
      "loss": 2.1186,
      "step": 486
    },
    {
      "epoch": 0.22329206785878036,
      "grad_norm": 0.22571653127670288,
      "learning_rate": 0.0009053797992469461,
      "loss": 2.1439,
      "step": 487
    },
    {
      "epoch": 0.22375057313159102,
      "grad_norm": 0.22372479736804962,
      "learning_rate": 0.0009049445949466078,
      "loss": 2.1429,
      "step": 488
    },
    {
      "epoch": 0.22420907840440166,
      "grad_norm": 0.23700931668281555,
      "learning_rate": 0.0009045084971874737,
      "loss": 2.1297,
      "step": 489
    },
    {
      "epoch": 0.2246675836772123,
      "grad_norm": 0.2604145407676697,
      "learning_rate": 0.0009040715069317382,
      "loss": 2.0468,
      "step": 490
    },
    {
      "epoch": 0.22512608895002292,
      "grad_norm": 0.24211353063583374,
      "learning_rate": 0.0009036336251435648,
      "loss": 2.1368,
      "step": 491
    },
    {
      "epoch": 0.22558459422283356,
      "grad_norm": 0.22774332761764526,
      "learning_rate": 0.0009031948527890839,
      "loss": 2.0919,
      "step": 492
    },
    {
      "epoch": 0.2260430994956442,
      "grad_norm": 0.26098906993865967,
      "learning_rate": 0.000902755190836391,
      "loss": 2.1603,
      "step": 493
    },
    {
      "epoch": 0.22650160476845485,
      "grad_norm": 0.24428172409534454,
      "learning_rate": 0.0009023146402555442,
      "loss": 2.1745,
      "step": 494
    },
    {
      "epoch": 0.22696011004126548,
      "grad_norm": 0.2387840449810028,
      "learning_rate": 0.0009018732020185624,
      "loss": 2.0865,
      "step": 495
    },
    {
      "epoch": 0.22741861531407612,
      "grad_norm": 0.2217601239681244,
      "learning_rate": 0.0009014308770994235,
      "loss": 2.1211,
      "step": 496
    },
    {
      "epoch": 0.22787712058688675,
      "grad_norm": 0.25421616435050964,
      "learning_rate": 0.0009009876664740605,
      "loss": 2.2144,
      "step": 497
    },
    {
      "epoch": 0.22833562585969738,
      "grad_norm": 0.24033181369304657,
      "learning_rate": 0.0009005435711203618,
      "loss": 2.2004,
      "step": 498
    },
    {
      "epoch": 0.22879413113250802,
      "grad_norm": 0.23308423161506653,
      "learning_rate": 0.000900098592018167,
      "loss": 2.1456,
      "step": 499
    },
    {
      "epoch": 0.22925263640531865,
      "grad_norm": 0.2245311439037323,
      "learning_rate": 0.0008996527301492663,
      "loss": 2.0852,
      "step": 500
    },
    {
      "epoch": 0.2297111416781293,
      "grad_norm": 0.2295488715171814,
      "learning_rate": 0.0008992059864973972,
      "loss": 2.0942,
      "step": 501
    },
    {
      "epoch": 0.23016964695093994,
      "grad_norm": 0.23744265735149384,
      "learning_rate": 0.0008987583620482427,
      "loss": 2.0366,
      "step": 502
    },
    {
      "epoch": 0.23062815222375058,
      "grad_norm": 0.23540134727954865,
      "learning_rate": 0.0008983098577894292,
      "loss": 2.1205,
      "step": 503
    },
    {
      "epoch": 0.2310866574965612,
      "grad_norm": 0.23612181842327118,
      "learning_rate": 0.0008978604747105246,
      "loss": 2.102,
      "step": 504
    },
    {
      "epoch": 0.23154516276937184,
      "grad_norm": 0.2557584047317505,
      "learning_rate": 0.0008974102138030354,
      "loss": 2.2768,
      "step": 505
    },
    {
      "epoch": 0.23200366804218248,
      "grad_norm": 0.2138661891222,
      "learning_rate": 0.000896959076060405,
      "loss": 2.1032,
      "step": 506
    },
    {
      "epoch": 0.2324621733149931,
      "grad_norm": 0.22726605832576752,
      "learning_rate": 0.0008965070624780116,
      "loss": 2.1368,
      "step": 507
    },
    {
      "epoch": 0.23292067858780377,
      "grad_norm": 0.22669675946235657,
      "learning_rate": 0.0008960541740531658,
      "loss": 2.0877,
      "step": 508
    },
    {
      "epoch": 0.2333791838606144,
      "grad_norm": 0.22224371135234833,
      "learning_rate": 0.0008956004117851083,
      "loss": 2.0829,
      "step": 509
    },
    {
      "epoch": 0.23383768913342504,
      "grad_norm": 0.23341889679431915,
      "learning_rate": 0.0008951457766750079,
      "loss": 2.2034,
      "step": 510
    },
    {
      "epoch": 0.23429619440623567,
      "grad_norm": 0.2240760177373886,
      "learning_rate": 0.0008946902697259593,
      "loss": 2.0779,
      "step": 511
    },
    {
      "epoch": 0.2347546996790463,
      "grad_norm": 0.21852333843708038,
      "learning_rate": 0.0008942338919429805,
      "loss": 2.1266,
      "step": 512
    },
    {
      "epoch": 0.23521320495185694,
      "grad_norm": 0.2283330112695694,
      "learning_rate": 0.0008937766443330113,
      "loss": 2.1407,
      "step": 513
    },
    {
      "epoch": 0.2356717102246676,
      "grad_norm": 0.22739507257938385,
      "learning_rate": 0.0008933185279049103,
      "loss": 2.1314,
      "step": 514
    },
    {
      "epoch": 0.23613021549747823,
      "grad_norm": 0.22821252048015594,
      "learning_rate": 0.0008928595436694532,
      "loss": 2.0913,
      "step": 515
    },
    {
      "epoch": 0.23658872077028886,
      "grad_norm": 0.2350698560476303,
      "learning_rate": 0.0008923996926393305,
      "loss": 2.1423,
      "step": 516
    },
    {
      "epoch": 0.2370472260430995,
      "grad_norm": 0.219473198056221,
      "learning_rate": 0.0008919389758291449,
      "loss": 2.0192,
      "step": 517
    },
    {
      "epoch": 0.23750573131591013,
      "grad_norm": 0.2161562442779541,
      "learning_rate": 0.0008914773942554098,
      "loss": 2.1254,
      "step": 518
    },
    {
      "epoch": 0.23796423658872076,
      "grad_norm": 0.20425254106521606,
      "learning_rate": 0.000891014948936546,
      "loss": 2.118,
      "step": 519
    },
    {
      "epoch": 0.2384227418615314,
      "grad_norm": 0.24168744683265686,
      "learning_rate": 0.0008905516408928804,
      "loss": 2.1572,
      "step": 520
    },
    {
      "epoch": 0.23888124713434206,
      "grad_norm": 0.2429639846086502,
      "learning_rate": 0.0008900874711466434,
      "loss": 2.2028,
      "step": 521
    },
    {
      "epoch": 0.2393397524071527,
      "grad_norm": 0.24031127989292145,
      "learning_rate": 0.0008896224407219666,
      "loss": 2.0751,
      "step": 522
    },
    {
      "epoch": 0.23979825767996332,
      "grad_norm": 0.2271275818347931,
      "learning_rate": 0.0008891565506448804,
      "loss": 2.0787,
      "step": 523
    },
    {
      "epoch": 0.24025676295277396,
      "grad_norm": 0.22774386405944824,
      "learning_rate": 0.0008886898019433122,
      "loss": 2.0622,
      "step": 524
    },
    {
      "epoch": 0.2407152682255846,
      "grad_norm": 0.2479386031627655,
      "learning_rate": 0.0008882221956470836,
      "loss": 2.1552,
      "step": 525
    },
    {
      "epoch": 0.24117377349839522,
      "grad_norm": 0.25274658203125,
      "learning_rate": 0.0008877537327879086,
      "loss": 2.1099,
      "step": 526
    },
    {
      "epoch": 0.24163227877120586,
      "grad_norm": 0.22654864192008972,
      "learning_rate": 0.0008872844143993908,
      "loss": 2.0674,
      "step": 527
    },
    {
      "epoch": 0.24209078404401652,
      "grad_norm": 0.22348535060882568,
      "learning_rate": 0.0008868142415170218,
      "loss": 2.0753,
      "step": 528
    },
    {
      "epoch": 0.24254928931682715,
      "grad_norm": 0.21471934020519257,
      "learning_rate": 0.0008863432151781781,
      "loss": 2.156,
      "step": 529
    },
    {
      "epoch": 0.24300779458963778,
      "grad_norm": 0.21842457354068756,
      "learning_rate": 0.0008858713364221195,
      "loss": 2.1238,
      "step": 530
    },
    {
      "epoch": 0.24346629986244842,
      "grad_norm": 0.21429242193698883,
      "learning_rate": 0.0008853986062899868,
      "loss": 2.1254,
      "step": 531
    },
    {
      "epoch": 0.24392480513525905,
      "grad_norm": 0.2091848999261856,
      "learning_rate": 0.0008849250258247986,
      "loss": 2.1264,
      "step": 532
    },
    {
      "epoch": 0.24438331040806968,
      "grad_norm": 0.21723726391792297,
      "learning_rate": 0.0008844505960714503,
      "loss": 2.0498,
      "step": 533
    },
    {
      "epoch": 0.24484181568088034,
      "grad_norm": 0.22974345088005066,
      "learning_rate": 0.0008839753180767108,
      "loss": 2.131,
      "step": 534
    },
    {
      "epoch": 0.24530032095369098,
      "grad_norm": 0.2293003797531128,
      "learning_rate": 0.0008834991928892204,
      "loss": 2.1938,
      "step": 535
    },
    {
      "epoch": 0.2457588262265016,
      "grad_norm": 0.23492081463336945,
      "learning_rate": 0.000883022221559489,
      "loss": 2.1397,
      "step": 536
    },
    {
      "epoch": 0.24621733149931224,
      "grad_norm": 0.24561959505081177,
      "learning_rate": 0.0008825444051398934,
      "loss": 2.1826,
      "step": 537
    },
    {
      "epoch": 0.24667583677212288,
      "grad_norm": 0.22359225153923035,
      "learning_rate": 0.0008820657446846745,
      "loss": 2.1652,
      "step": 538
    },
    {
      "epoch": 0.2471343420449335,
      "grad_norm": 0.2149883359670639,
      "learning_rate": 0.000881586241249936,
      "loss": 2.0281,
      "step": 539
    },
    {
      "epoch": 0.24759284731774414,
      "grad_norm": 0.22548729181289673,
      "learning_rate": 0.0008811058958936411,
      "loss": 2.1396,
      "step": 540
    },
    {
      "epoch": 0.2480513525905548,
      "grad_norm": 0.21922555565834045,
      "learning_rate": 0.000880624709675611,
      "loss": 2.0344,
      "step": 541
    },
    {
      "epoch": 0.24850985786336544,
      "grad_norm": 0.2166142612695694,
      "learning_rate": 0.000880142683657522,
      "loss": 2.1667,
      "step": 542
    },
    {
      "epoch": 0.24896836313617607,
      "grad_norm": 0.21259035170078278,
      "learning_rate": 0.0008796598189029029,
      "loss": 2.0823,
      "step": 543
    },
    {
      "epoch": 0.2494268684089867,
      "grad_norm": 0.21874083578586578,
      "learning_rate": 0.0008791761164771338,
      "loss": 2.0649,
      "step": 544
    },
    {
      "epoch": 0.24988537368179733,
      "grad_norm": 0.23225602507591248,
      "learning_rate": 0.0008786915774474424,
      "loss": 2.1374,
      "step": 545
    },
    {
      "epoch": 0.25034387895460797,
      "grad_norm": 0.241831973195076,
      "learning_rate": 0.0008782062028829027,
      "loss": 2.2109,
      "step": 546
    },
    {
      "epoch": 0.25080238422741863,
      "grad_norm": 0.22860047221183777,
      "learning_rate": 0.0008777199938544318,
      "loss": 2.0887,
      "step": 547
    },
    {
      "epoch": 0.25126088950022923,
      "grad_norm": 0.22134608030319214,
      "learning_rate": 0.0008772329514347883,
      "loss": 2.0716,
      "step": 548
    },
    {
      "epoch": 0.2517193947730399,
      "grad_norm": 0.2337459772825241,
      "learning_rate": 0.0008767450766985694,
      "loss": 2.2061,
      "step": 549
    },
    {
      "epoch": 0.2521779000458505,
      "grad_norm": 0.2120031863451004,
      "learning_rate": 0.0008762563707222086,
      "loss": 2.0902,
      "step": 550
    },
    {
      "epoch": 0.25263640531866116,
      "grad_norm": 0.21332746744155884,
      "learning_rate": 0.0008757668345839738,
      "loss": 2.1372,
      "step": 551
    },
    {
      "epoch": 0.2530949105914718,
      "grad_norm": 0.2189939171075821,
      "learning_rate": 0.0008752764693639638,
      "loss": 2.108,
      "step": 552
    },
    {
      "epoch": 0.2535534158642824,
      "grad_norm": 0.22233176231384277,
      "learning_rate": 0.0008747852761441078,
      "loss": 2.2069,
      "step": 553
    },
    {
      "epoch": 0.2540119211370931,
      "grad_norm": 0.22609347105026245,
      "learning_rate": 0.0008742932560081607,
      "loss": 2.121,
      "step": 554
    },
    {
      "epoch": 0.2544704264099037,
      "grad_norm": 0.21771842241287231,
      "learning_rate": 0.0008738004100417025,
      "loss": 2.1471,
      "step": 555
    },
    {
      "epoch": 0.25492893168271435,
      "grad_norm": 0.22729527950286865,
      "learning_rate": 0.0008733067393321355,
      "loss": 2.138,
      "step": 556
    },
    {
      "epoch": 0.255387436955525,
      "grad_norm": 0.2119305282831192,
      "learning_rate": 0.000872812244968681,
      "loss": 2.144,
      "step": 557
    },
    {
      "epoch": 0.2558459422283356,
      "grad_norm": 0.20920993387699127,
      "learning_rate": 0.0008723169280423783,
      "loss": 2.1409,
      "step": 558
    },
    {
      "epoch": 0.2563044475011463,
      "grad_norm": 0.2103026658296585,
      "learning_rate": 0.0008718207896460811,
      "loss": 2.1275,
      "step": 559
    },
    {
      "epoch": 0.2567629527739569,
      "grad_norm": 0.21533221006393433,
      "learning_rate": 0.0008713238308744557,
      "loss": 2.0805,
      "step": 560
    },
    {
      "epoch": 0.25722145804676755,
      "grad_norm": 0.22689302265644073,
      "learning_rate": 0.0008708260528239789,
      "loss": 2.0733,
      "step": 561
    },
    {
      "epoch": 0.25767996331957815,
      "grad_norm": 0.21749329566955566,
      "learning_rate": 0.000870327456592934,
      "loss": 2.0817,
      "step": 562
    },
    {
      "epoch": 0.2581384685923888,
      "grad_norm": 0.2315261960029602,
      "learning_rate": 0.0008698280432814107,
      "loss": 2.1259,
      "step": 563
    },
    {
      "epoch": 0.2585969738651995,
      "grad_norm": 0.22803950309753418,
      "learning_rate": 0.000869327813991301,
      "loss": 2.1627,
      "step": 564
    },
    {
      "epoch": 0.2590554791380101,
      "grad_norm": 0.23358319699764252,
      "learning_rate": 0.0008688267698262971,
      "loss": 2.1521,
      "step": 565
    },
    {
      "epoch": 0.25951398441082074,
      "grad_norm": 0.23222728073596954,
      "learning_rate": 0.0008683249118918894,
      "loss": 2.149,
      "step": 566
    },
    {
      "epoch": 0.25997248968363135,
      "grad_norm": 0.25268107652664185,
      "learning_rate": 0.0008678222412953637,
      "loss": 2.1271,
      "step": 567
    },
    {
      "epoch": 0.260430994956442,
      "grad_norm": 0.22046995162963867,
      "learning_rate": 0.0008673187591457987,
      "loss": 2.2304,
      "step": 568
    },
    {
      "epoch": 0.2608895002292526,
      "grad_norm": 0.24098508059978485,
      "learning_rate": 0.0008668144665540639,
      "loss": 2.1554,
      "step": 569
    },
    {
      "epoch": 0.2613480055020633,
      "grad_norm": 0.2123943269252777,
      "learning_rate": 0.0008663093646328167,
      "loss": 2.1487,
      "step": 570
    },
    {
      "epoch": 0.26180651077487394,
      "grad_norm": 0.22088319063186646,
      "learning_rate": 0.0008658034544965003,
      "loss": 2.0185,
      "step": 571
    },
    {
      "epoch": 0.26226501604768454,
      "grad_norm": 0.2234388142824173,
      "learning_rate": 0.0008652967372613412,
      "loss": 2.1204,
      "step": 572
    },
    {
      "epoch": 0.2627235213204952,
      "grad_norm": 0.2150997519493103,
      "learning_rate": 0.0008647892140453466,
      "loss": 2.0351,
      "step": 573
    },
    {
      "epoch": 0.2631820265933058,
      "grad_norm": 0.214493528008461,
      "learning_rate": 0.0008642808859683021,
      "loss": 2.0354,
      "step": 574
    },
    {
      "epoch": 0.26364053186611647,
      "grad_norm": 0.2791043519973755,
      "learning_rate": 0.0008637717541517689,
      "loss": 2.1731,
      "step": 575
    },
    {
      "epoch": 0.2640990371389271,
      "grad_norm": 0.2206440418958664,
      "learning_rate": 0.0008632618197190816,
      "loss": 2.1354,
      "step": 576
    },
    {
      "epoch": 0.26455754241173773,
      "grad_norm": 0.2407679259777069,
      "learning_rate": 0.0008627510837953458,
      "loss": 2.1633,
      "step": 577
    },
    {
      "epoch": 0.2650160476845484,
      "grad_norm": 0.22174328565597534,
      "learning_rate": 0.0008622395475074355,
      "loss": 2.1388,
      "step": 578
    },
    {
      "epoch": 0.265474552957359,
      "grad_norm": 0.23405641317367554,
      "learning_rate": 0.0008617272119839903,
      "loss": 2.1023,
      "step": 579
    },
    {
      "epoch": 0.26593305823016966,
      "grad_norm": 0.2251880019903183,
      "learning_rate": 0.0008612140783554136,
      "loss": 2.2015,
      "step": 580
    },
    {
      "epoch": 0.26639156350298027,
      "grad_norm": 0.23050104081630707,
      "learning_rate": 0.0008607001477538696,
      "loss": 2.102,
      "step": 581
    },
    {
      "epoch": 0.2668500687757909,
      "grad_norm": 0.23793485760688782,
      "learning_rate": 0.0008601854213132807,
      "loss": 2.1785,
      "step": 582
    },
    {
      "epoch": 0.26730857404860153,
      "grad_norm": 0.24545785784721375,
      "learning_rate": 0.0008596699001693256,
      "loss": 2.1225,
      "step": 583
    },
    {
      "epoch": 0.2677670793214122,
      "grad_norm": 0.23357443511486053,
      "learning_rate": 0.000859153585459436,
      "loss": 2.1256,
      "step": 584
    },
    {
      "epoch": 0.26822558459422285,
      "grad_norm": 0.2164115607738495,
      "learning_rate": 0.0008586364783227949,
      "loss": 2.1557,
      "step": 585
    },
    {
      "epoch": 0.26868408986703346,
      "grad_norm": 0.24592041969299316,
      "learning_rate": 0.0008581185799003332,
      "loss": 2.1599,
      "step": 586
    },
    {
      "epoch": 0.2691425951398441,
      "grad_norm": 0.2248750478029251,
      "learning_rate": 0.0008575998913347283,
      "loss": 2.2013,
      "step": 587
    },
    {
      "epoch": 0.2696011004126547,
      "grad_norm": 0.21157236397266388,
      "learning_rate": 0.0008570804137704004,
      "loss": 2.1095,
      "step": 588
    },
    {
      "epoch": 0.2700596056854654,
      "grad_norm": 0.23289994895458221,
      "learning_rate": 0.0008565601483535108,
      "loss": 2.0713,
      "step": 589
    },
    {
      "epoch": 0.27051811095827605,
      "grad_norm": 0.2292129546403885,
      "learning_rate": 0.0008560390962319591,
      "loss": 2.1341,
      "step": 590
    },
    {
      "epoch": 0.27097661623108665,
      "grad_norm": 0.2293434739112854,
      "learning_rate": 0.0008555172585553804,
      "loss": 2.1117,
      "step": 591
    },
    {
      "epoch": 0.2714351215038973,
      "grad_norm": 0.2139606475830078,
      "learning_rate": 0.0008549946364751435,
      "loss": 2.0672,
      "step": 592
    },
    {
      "epoch": 0.2718936267767079,
      "grad_norm": 0.22785142064094543,
      "learning_rate": 0.0008544712311443475,
      "loss": 2.1425,
      "step": 593
    },
    {
      "epoch": 0.2723521320495186,
      "grad_norm": 0.24359388649463654,
      "learning_rate": 0.0008539470437178196,
      "loss": 2.1282,
      "step": 594
    },
    {
      "epoch": 0.2728106373223292,
      "grad_norm": 0.20478704571723938,
      "learning_rate": 0.000853422075352113,
      "loss": 2.1015,
      "step": 595
    },
    {
      "epoch": 0.27326914259513985,
      "grad_norm": 0.21177560091018677,
      "learning_rate": 0.0008528963272055035,
      "loss": 2.1727,
      "step": 596
    },
    {
      "epoch": 0.2737276478679505,
      "grad_norm": 0.20961932837963104,
      "learning_rate": 0.0008523698004379877,
      "loss": 2.1137,
      "step": 597
    },
    {
      "epoch": 0.2741861531407611,
      "grad_norm": 0.22459998726844788,
      "learning_rate": 0.00085184249621128,
      "loss": 2.0698,
      "step": 598
    },
    {
      "epoch": 0.2746446584135718,
      "grad_norm": 0.2203604280948639,
      "learning_rate": 0.0008513144156888101,
      "loss": 2.1351,
      "step": 599
    },
    {
      "epoch": 0.2751031636863824,
      "grad_norm": 0.22351498901844025,
      "learning_rate": 0.0008507855600357207,
      "loss": 2.1951,
      "step": 600
    },
    {
      "epoch": 0.27556166895919304,
      "grad_norm": 0.21693259477615356,
      "learning_rate": 0.0008502559304188644,
      "loss": 2.0759,
      "step": 601
    },
    {
      "epoch": 0.27602017423200365,
      "grad_norm": 0.2427276372909546,
      "learning_rate": 0.0008497255280068019,
      "loss": 2.1518,
      "step": 602
    },
    {
      "epoch": 0.2764786795048143,
      "grad_norm": 0.23979704082012177,
      "learning_rate": 0.0008491943539697986,
      "loss": 2.1138,
      "step": 603
    },
    {
      "epoch": 0.27693718477762497,
      "grad_norm": 0.20996230840682983,
      "learning_rate": 0.0008486624094798226,
      "loss": 2.0837,
      "step": 604
    },
    {
      "epoch": 0.2773956900504356,
      "grad_norm": 0.2272711992263794,
      "learning_rate": 0.0008481296957105417,
      "loss": 2.1204,
      "step": 605
    },
    {
      "epoch": 0.27785419532324623,
      "grad_norm": 0.22825433313846588,
      "learning_rate": 0.0008475962138373213,
      "loss": 2.1512,
      "step": 606
    },
    {
      "epoch": 0.27831270059605684,
      "grad_norm": 0.23122191429138184,
      "learning_rate": 0.0008470619650372211,
      "loss": 2.0754,
      "step": 607
    },
    {
      "epoch": 0.2787712058688675,
      "grad_norm": 0.22791150212287903,
      "learning_rate": 0.0008465269504889934,
      "loss": 2.1075,
      "step": 608
    },
    {
      "epoch": 0.2792297111416781,
      "grad_norm": 0.20903818309307098,
      "learning_rate": 0.0008459911713730799,
      "loss": 2.1466,
      "step": 609
    },
    {
      "epoch": 0.27968821641448877,
      "grad_norm": 0.2086869776248932,
      "learning_rate": 0.0008454546288716089,
      "loss": 2.0978,
      "step": 610
    },
    {
      "epoch": 0.2801467216872994,
      "grad_norm": 0.22324064373970032,
      "learning_rate": 0.0008449173241683935,
      "loss": 2.0902,
      "step": 611
    },
    {
      "epoch": 0.28060522696011003,
      "grad_norm": 0.2228441685438156,
      "learning_rate": 0.0008443792584489281,
      "loss": 2.1909,
      "step": 612
    },
    {
      "epoch": 0.2810637322329207,
      "grad_norm": 0.21778002381324768,
      "learning_rate": 0.0008438404329003863,
      "loss": 2.1199,
      "step": 613
    },
    {
      "epoch": 0.2815222375057313,
      "grad_norm": 0.22855491936206818,
      "learning_rate": 0.0008433008487116183,
      "loss": 2.0704,
      "step": 614
    },
    {
      "epoch": 0.28198074277854196,
      "grad_norm": 0.22671599686145782,
      "learning_rate": 0.0008427605070731481,
      "loss": 2.0718,
      "step": 615
    },
    {
      "epoch": 0.28243924805135256,
      "grad_norm": 0.22949114441871643,
      "learning_rate": 0.0008422194091771708,
      "loss": 2.1059,
      "step": 616
    },
    {
      "epoch": 0.2828977533241632,
      "grad_norm": 0.22040203213691711,
      "learning_rate": 0.0008416775562175503,
      "loss": 2.1061,
      "step": 617
    },
    {
      "epoch": 0.2833562585969739,
      "grad_norm": 0.2124480903148651,
      "learning_rate": 0.000841134949389816,
      "loss": 2.0931,
      "step": 618
    },
    {
      "epoch": 0.2838147638697845,
      "grad_norm": 0.21799595654010773,
      "learning_rate": 0.0008405915898911611,
      "loss": 2.1379,
      "step": 619
    },
    {
      "epoch": 0.28427326914259515,
      "grad_norm": 0.21503432095050812,
      "learning_rate": 0.0008400474789204396,
      "loss": 2.1272,
      "step": 620
    },
    {
      "epoch": 0.28473177441540576,
      "grad_norm": 0.20371070504188538,
      "learning_rate": 0.0008395026176781626,
      "loss": 2.0894,
      "step": 621
    },
    {
      "epoch": 0.2851902796882164,
      "grad_norm": 0.2199876755475998,
      "learning_rate": 0.0008389570073664976,
      "loss": 2.1393,
      "step": 622
    },
    {
      "epoch": 0.285648784961027,
      "grad_norm": 0.22322788834571838,
      "learning_rate": 0.0008384106491892642,
      "loss": 2.1829,
      "step": 623
    },
    {
      "epoch": 0.2861072902338377,
      "grad_norm": 0.20997585356235504,
      "learning_rate": 0.0008378635443519327,
      "loss": 2.124,
      "step": 624
    },
    {
      "epoch": 0.28656579550664835,
      "grad_norm": 0.2175171971321106,
      "learning_rate": 0.0008373156940616199,
      "loss": 2.1043,
      "step": 625
    },
    {
      "epoch": 0.28702430077945895,
      "grad_norm": 0.21799927949905396,
      "learning_rate": 0.0008367670995270882,
      "loss": 2.1579,
      "step": 626
    },
    {
      "epoch": 0.2874828060522696,
      "grad_norm": 0.22054778039455414,
      "learning_rate": 0.0008362177619587416,
      "loss": 2.0714,
      "step": 627
    },
    {
      "epoch": 0.2879413113250802,
      "grad_norm": 0.213394433259964,
      "learning_rate": 0.0008356676825686238,
      "loss": 2.0598,
      "step": 628
    },
    {
      "epoch": 0.2883998165978909,
      "grad_norm": 0.22988812625408173,
      "learning_rate": 0.0008351168625704147,
      "loss": 2.2235,
      "step": 629
    },
    {
      "epoch": 0.28885832187070154,
      "grad_norm": 0.22170864045619965,
      "learning_rate": 0.0008345653031794292,
      "loss": 2.1114,
      "step": 630
    },
    {
      "epoch": 0.28931682714351215,
      "grad_norm": 0.20631034672260284,
      "learning_rate": 0.0008340130056126125,
      "loss": 2.0231,
      "step": 631
    },
    {
      "epoch": 0.2897753324163228,
      "grad_norm": 0.21015353500843048,
      "learning_rate": 0.0008334599710885394,
      "loss": 2.1079,
      "step": 632
    },
    {
      "epoch": 0.2902338376891334,
      "grad_norm": 0.2207147181034088,
      "learning_rate": 0.0008329062008274098,
      "loss": 2.0268,
      "step": 633
    },
    {
      "epoch": 0.2906923429619441,
      "grad_norm": 0.20542165637016296,
      "learning_rate": 0.000832351696051048,
      "loss": 2.0865,
      "step": 634
    },
    {
      "epoch": 0.2911508482347547,
      "grad_norm": 0.20227493345737457,
      "learning_rate": 0.000831796457982898,
      "loss": 2.1187,
      "step": 635
    },
    {
      "epoch": 0.29160935350756534,
      "grad_norm": 0.2174948900938034,
      "learning_rate": 0.0008312404878480222,
      "loss": 2.1279,
      "step": 636
    },
    {
      "epoch": 0.292067858780376,
      "grad_norm": 0.2821820080280304,
      "learning_rate": 0.0008306837868730979,
      "loss": 2.1181,
      "step": 637
    },
    {
      "epoch": 0.2925263640531866,
      "grad_norm": 0.2283606231212616,
      "learning_rate": 0.0008301263562864152,
      "loss": 2.1588,
      "step": 638
    },
    {
      "epoch": 0.29298486932599727,
      "grad_norm": 0.2315107136964798,
      "learning_rate": 0.0008295681973178737,
      "loss": 2.1751,
      "step": 639
    },
    {
      "epoch": 0.29344337459880787,
      "grad_norm": 0.22781269252300262,
      "learning_rate": 0.0008290093111989804,
      "loss": 2.0983,
      "step": 640
    },
    {
      "epoch": 0.29390187987161853,
      "grad_norm": 0.24152600765228271,
      "learning_rate": 0.0008284496991628465,
      "loss": 2.0625,
      "step": 641
    },
    {
      "epoch": 0.29436038514442914,
      "grad_norm": 0.224074125289917,
      "learning_rate": 0.0008278893624441847,
      "loss": 2.1555,
      "step": 642
    },
    {
      "epoch": 0.2948188904172398,
      "grad_norm": 0.2265140563249588,
      "learning_rate": 0.000827328302279307,
      "loss": 2.0499,
      "step": 643
    },
    {
      "epoch": 0.29527739569005046,
      "grad_norm": 0.21782460808753967,
      "learning_rate": 0.0008267665199061211,
      "loss": 2.0972,
      "step": 644
    },
    {
      "epoch": 0.29573590096286106,
      "grad_norm": 0.23569810390472412,
      "learning_rate": 0.0008262040165641288,
      "loss": 2.1303,
      "step": 645
    },
    {
      "epoch": 0.2961944062356717,
      "grad_norm": 0.2238384187221527,
      "learning_rate": 0.0008256407934944219,
      "loss": 2.1309,
      "step": 646
    },
    {
      "epoch": 0.29665291150848233,
      "grad_norm": 0.22050710022449493,
      "learning_rate": 0.0008250768519396807,
      "loss": 2.044,
      "step": 647
    },
    {
      "epoch": 0.297111416781293,
      "grad_norm": 0.20564749836921692,
      "learning_rate": 0.0008245121931441706,
      "loss": 2.0925,
      "step": 648
    },
    {
      "epoch": 0.2975699220541036,
      "grad_norm": 0.2131083905696869,
      "learning_rate": 0.0008239468183537393,
      "loss": 2.1941,
      "step": 649
    },
    {
      "epoch": 0.29802842732691426,
      "grad_norm": 0.23625297844409943,
      "learning_rate": 0.0008233807288158146,
      "loss": 2.0953,
      "step": 650
    },
    {
      "epoch": 0.2984869325997249,
      "grad_norm": 0.20378124713897705,
      "learning_rate": 0.0008228139257794012,
      "loss": 2.1179,
      "step": 651
    },
    {
      "epoch": 0.2989454378725355,
      "grad_norm": 0.22629019618034363,
      "learning_rate": 0.0008222464104950778,
      "loss": 2.1374,
      "step": 652
    },
    {
      "epoch": 0.2994039431453462,
      "grad_norm": 0.21908162534236908,
      "learning_rate": 0.000821678184214995,
      "loss": 2.1189,
      "step": 653
    },
    {
      "epoch": 0.2998624484181568,
      "grad_norm": 0.22726264595985413,
      "learning_rate": 0.0008211092481928716,
      "loss": 2.0653,
      "step": 654
    },
    {
      "epoch": 0.30032095369096745,
      "grad_norm": 0.2398594468832016,
      "learning_rate": 0.0008205396036839927,
      "loss": 2.1336,
      "step": 655
    },
    {
      "epoch": 0.30077945896377806,
      "grad_norm": 0.22590592503547668,
      "learning_rate": 0.0008199692519452069,
      "loss": 2.2114,
      "step": 656
    },
    {
      "epoch": 0.3012379642365887,
      "grad_norm": 0.23581646382808685,
      "learning_rate": 0.0008193981942349224,
      "loss": 2.0639,
      "step": 657
    },
    {
      "epoch": 0.3016964695093994,
      "grad_norm": 0.23422463238239288,
      "learning_rate": 0.0008188264318131056,
      "loss": 2.0986,
      "step": 658
    },
    {
      "epoch": 0.30215497478221,
      "grad_norm": 0.24346496164798737,
      "learning_rate": 0.0008182539659412776,
      "loss": 2.0751,
      "step": 659
    },
    {
      "epoch": 0.30261348005502064,
      "grad_norm": 0.2397485375404358,
      "learning_rate": 0.0008176807978825118,
      "loss": 2.0806,
      "step": 660
    },
    {
      "epoch": 0.30307198532783125,
      "grad_norm": 0.23259598016738892,
      "learning_rate": 0.0008171069289014306,
      "loss": 2.0392,
      "step": 661
    },
    {
      "epoch": 0.3035304906006419,
      "grad_norm": 0.22700823843479156,
      "learning_rate": 0.0008165323602642028,
      "loss": 2.1772,
      "step": 662
    },
    {
      "epoch": 0.30398899587345257,
      "grad_norm": 0.22056011855602264,
      "learning_rate": 0.0008159570932385414,
      "loss": 2.1411,
      "step": 663
    },
    {
      "epoch": 0.3044475011462632,
      "grad_norm": 0.22846128046512604,
      "learning_rate": 0.0008153811290936999,
      "loss": 2.1604,
      "step": 664
    },
    {
      "epoch": 0.30490600641907384,
      "grad_norm": 0.22465088963508606,
      "learning_rate": 0.0008148044691004698,
      "loss": 2.144,
      "step": 665
    },
    {
      "epoch": 0.30536451169188444,
      "grad_norm": 0.21845774352550507,
      "learning_rate": 0.0008142271145311783,
      "loss": 2.1131,
      "step": 666
    },
    {
      "epoch": 0.3058230169646951,
      "grad_norm": 0.2276507169008255,
      "learning_rate": 0.000813649066659685,
      "loss": 2.1198,
      "step": 667
    },
    {
      "epoch": 0.3062815222375057,
      "grad_norm": 0.21931704878807068,
      "learning_rate": 0.0008130703267613787,
      "loss": 2.0899,
      "step": 668
    },
    {
      "epoch": 0.30674002751031637,
      "grad_norm": 0.24375027418136597,
      "learning_rate": 0.0008124908961131759,
      "loss": 2.101,
      "step": 669
    },
    {
      "epoch": 0.30719853278312703,
      "grad_norm": 0.23315945267677307,
      "learning_rate": 0.0008119107759935163,
      "loss": 2.0851,
      "step": 670
    },
    {
      "epoch": 0.30765703805593764,
      "grad_norm": 0.21995766460895538,
      "learning_rate": 0.0008113299676823615,
      "loss": 2.1313,
      "step": 671
    },
    {
      "epoch": 0.3081155433287483,
      "grad_norm": 0.24809050559997559,
      "learning_rate": 0.0008107484724611911,
      "loss": 2.1015,
      "step": 672
    },
    {
      "epoch": 0.3085740486015589,
      "grad_norm": 0.21844452619552612,
      "learning_rate": 0.0008101662916130006,
      "loss": 2.1089,
      "step": 673
    },
    {
      "epoch": 0.30903255387436956,
      "grad_norm": 0.21229973435401917,
      "learning_rate": 0.0008095834264222979,
      "loss": 2.1184,
      "step": 674
    },
    {
      "epoch": 0.30949105914718017,
      "grad_norm": 0.2036106139421463,
      "learning_rate": 0.0008089998781751009,
      "loss": 2.2374,
      "step": 675
    },
    {
      "epoch": 0.30994956441999083,
      "grad_norm": 0.2052416205406189,
      "learning_rate": 0.0008084156481589349,
      "loss": 2.1654,
      "step": 676
    },
    {
      "epoch": 0.3104080696928015,
      "grad_norm": 0.2215793877840042,
      "learning_rate": 0.0008078307376628291,
      "loss": 2.1143,
      "step": 677
    },
    {
      "epoch": 0.3108665749656121,
      "grad_norm": 0.2110101580619812,
      "learning_rate": 0.0008072451479773143,
      "loss": 2.1906,
      "step": 678
    },
    {
      "epoch": 0.31132508023842276,
      "grad_norm": 0.20637603104114532,
      "learning_rate": 0.0008066588803944195,
      "loss": 2.0886,
      "step": 679
    },
    {
      "epoch": 0.31178358551123336,
      "grad_norm": 0.2247067093849182,
      "learning_rate": 0.0008060719362076697,
      "loss": 2.1618,
      "step": 680
    },
    {
      "epoch": 0.312242090784044,
      "grad_norm": 0.21900518238544464,
      "learning_rate": 0.0008054843167120826,
      "loss": 2.0733,
      "step": 681
    },
    {
      "epoch": 0.31270059605685463,
      "grad_norm": 0.22218984365463257,
      "learning_rate": 0.0008048960232041663,
      "loss": 2.1524,
      "step": 682
    },
    {
      "epoch": 0.3131591013296653,
      "grad_norm": 0.23978249728679657,
      "learning_rate": 0.0008043070569819153,
      "loss": 2.046,
      "step": 683
    },
    {
      "epoch": 0.31361760660247595,
      "grad_norm": 0.22147485613822937,
      "learning_rate": 0.0008037174193448089,
      "loss": 2.1495,
      "step": 684
    },
    {
      "epoch": 0.31407611187528656,
      "grad_norm": 0.22252248227596283,
      "learning_rate": 0.0008031271115938077,
      "loss": 2.1222,
      "step": 685
    },
    {
      "epoch": 0.3145346171480972,
      "grad_norm": 0.22622334957122803,
      "learning_rate": 0.0008025361350313505,
      "loss": 2.1255,
      "step": 686
    },
    {
      "epoch": 0.3149931224209078,
      "grad_norm": 0.21564321219921112,
      "learning_rate": 0.0008019444909613523,
      "loss": 2.1227,
      "step": 687
    },
    {
      "epoch": 0.3154516276937185,
      "grad_norm": 0.21427173912525177,
      "learning_rate": 0.0008013521806892003,
      "loss": 2.0057,
      "step": 688
    },
    {
      "epoch": 0.3159101329665291,
      "grad_norm": 0.2206098884344101,
      "learning_rate": 0.000800759205521752,
      "loss": 2.1459,
      "step": 689
    },
    {
      "epoch": 0.31636863823933975,
      "grad_norm": 0.21793878078460693,
      "learning_rate": 0.0008001655667673318,
      "loss": 2.1836,
      "step": 690
    },
    {
      "epoch": 0.3168271435121504,
      "grad_norm": 0.2118593454360962,
      "learning_rate": 0.0007995712657357279,
      "loss": 2.1338,
      "step": 691
    },
    {
      "epoch": 0.317285648784961,
      "grad_norm": 0.20529083907604218,
      "learning_rate": 0.0007989763037381904,
      "loss": 2.204,
      "step": 692
    },
    {
      "epoch": 0.3177441540577717,
      "grad_norm": 0.21344071626663208,
      "learning_rate": 0.0007983806820874271,
      "loss": 2.0803,
      "step": 693
    },
    {
      "epoch": 0.3182026593305823,
      "grad_norm": 0.21106135845184326,
      "learning_rate": 0.0007977844020976016,
      "loss": 2.0722,
      "step": 694
    },
    {
      "epoch": 0.31866116460339294,
      "grad_norm": 0.231025829911232,
      "learning_rate": 0.00079718746508433,
      "loss": 2.1485,
      "step": 695
    },
    {
      "epoch": 0.31911966987620355,
      "grad_norm": 0.2071506828069687,
      "learning_rate": 0.0007965898723646776,
      "loss": 2.134,
      "step": 696
    },
    {
      "epoch": 0.3195781751490142,
      "grad_norm": 0.20878209173679352,
      "learning_rate": 0.0007959916252571573,
      "loss": 2.0727,
      "step": 697
    },
    {
      "epoch": 0.32003668042182487,
      "grad_norm": 0.21293403208255768,
      "learning_rate": 0.000795392725081725,
      "loss": 2.046,
      "step": 698
    },
    {
      "epoch": 0.3204951856946355,
      "grad_norm": 0.2346765547990799,
      "learning_rate": 0.000794793173159778,
      "loss": 2.1671,
      "step": 699
    },
    {
      "epoch": 0.32095369096744614,
      "grad_norm": 0.222706139087677,
      "learning_rate": 0.0007941929708141513,
      "loss": 2.2033,
      "step": 700
    },
    {
      "epoch": 0.32141219624025674,
      "grad_norm": 0.2136891931295395,
      "learning_rate": 0.0007935921193691153,
      "loss": 2.0944,
      "step": 701
    },
    {
      "epoch": 0.3218707015130674,
      "grad_norm": 0.22754491865634918,
      "learning_rate": 0.0007929906201503722,
      "loss": 2.1051,
      "step": 702
    },
    {
      "epoch": 0.32232920678587806,
      "grad_norm": 0.2080554962158203,
      "learning_rate": 0.0007923884744850536,
      "loss": 2.1272,
      "step": 703
    },
    {
      "epoch": 0.32278771205868867,
      "grad_norm": 0.2290031760931015,
      "learning_rate": 0.0007917856837017176,
      "loss": 2.0787,
      "step": 704
    },
    {
      "epoch": 0.32324621733149933,
      "grad_norm": 0.2235427051782608,
      "learning_rate": 0.0007911822491303452,
      "loss": 2.1161,
      "step": 705
    },
    {
      "epoch": 0.32370472260430994,
      "grad_norm": 0.20676518976688385,
      "learning_rate": 0.0007905781721023382,
      "loss": 2.1074,
      "step": 706
    },
    {
      "epoch": 0.3241632278771206,
      "grad_norm": 0.22672808170318604,
      "learning_rate": 0.000789973453950516,
      "loss": 2.0502,
      "step": 707
    },
    {
      "epoch": 0.3246217331499312,
      "grad_norm": 0.21325217187404633,
      "learning_rate": 0.000789368096009112,
      "loss": 2.1219,
      "step": 708
    },
    {
      "epoch": 0.32508023842274186,
      "grad_norm": 0.22571179270744324,
      "learning_rate": 0.0007887620996137721,
      "loss": 2.073,
      "step": 709
    },
    {
      "epoch": 0.3255387436955525,
      "grad_norm": 0.2157471477985382,
      "learning_rate": 0.0007881554661015497,
      "loss": 2.038,
      "step": 710
    },
    {
      "epoch": 0.32599724896836313,
      "grad_norm": 0.22305913269519806,
      "learning_rate": 0.0007875481968109051,
      "loss": 2.0996,
      "step": 711
    },
    {
      "epoch": 0.3264557542411738,
      "grad_norm": 0.2123527079820633,
      "learning_rate": 0.0007869402930817007,
      "loss": 2.1512,
      "step": 712
    },
    {
      "epoch": 0.3269142595139844,
      "grad_norm": 0.2122092992067337,
      "learning_rate": 0.0007863317562551987,
      "loss": 2.0135,
      "step": 713
    },
    {
      "epoch": 0.32737276478679506,
      "grad_norm": 0.2227988988161087,
      "learning_rate": 0.0007857225876740584,
      "loss": 2.0936,
      "step": 714
    },
    {
      "epoch": 0.32783127005960566,
      "grad_norm": 0.21466785669326782,
      "learning_rate": 0.0007851127886823327,
      "loss": 2.0268,
      "step": 715
    },
    {
      "epoch": 0.3282897753324163,
      "grad_norm": 0.2051657736301422,
      "learning_rate": 0.0007845023606254658,
      "loss": 2.0788,
      "step": 716
    },
    {
      "epoch": 0.328748280605227,
      "grad_norm": 0.21836677193641663,
      "learning_rate": 0.0007838913048502894,
      "loss": 2.0623,
      "step": 717
    },
    {
      "epoch": 0.3292067858780376,
      "grad_norm": 0.2478582262992859,
      "learning_rate": 0.0007832796227050208,
      "loss": 2.2498,
      "step": 718
    },
    {
      "epoch": 0.32966529115084825,
      "grad_norm": 0.22686585783958435,
      "learning_rate": 0.0007826673155392587,
      "loss": 2.1115,
      "step": 719
    },
    {
      "epoch": 0.33012379642365886,
      "grad_norm": 0.21020585298538208,
      "learning_rate": 0.000782054384703981,
      "loss": 2.0377,
      "step": 720
    },
    {
      "epoch": 0.3305823016964695,
      "grad_norm": 0.21286536753177643,
      "learning_rate": 0.0007814408315515418,
      "loss": 2.0898,
      "step": 721
    },
    {
      "epoch": 0.3310408069692801,
      "grad_norm": 0.21350909769535065,
      "learning_rate": 0.0007808266574356683,
      "loss": 2.164,
      "step": 722
    },
    {
      "epoch": 0.3314993122420908,
      "grad_norm": 0.24148285388946533,
      "learning_rate": 0.0007802118637114573,
      "loss": 2.0766,
      "step": 723
    },
    {
      "epoch": 0.33195781751490144,
      "grad_norm": 0.2074369639158249,
      "learning_rate": 0.0007795964517353734,
      "loss": 2.1702,
      "step": 724
    },
    {
      "epoch": 0.33241632278771205,
      "grad_norm": 0.21709533035755157,
      "learning_rate": 0.0007789804228652449,
      "loss": 2.1106,
      "step": 725
    },
    {
      "epoch": 0.3328748280605227,
      "grad_norm": 0.20741873979568481,
      "learning_rate": 0.0007783637784602609,
      "loss": 2.0478,
      "step": 726
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.20994415879249573,
      "learning_rate": 0.0007777465198809692,
      "loss": 2.1079,
      "step": 727
    },
    {
      "epoch": 0.333791838606144,
      "grad_norm": 0.24055634438991547,
      "learning_rate": 0.0007771286484892722,
      "loss": 2.2053,
      "step": 728
    },
    {
      "epoch": 0.3342503438789546,
      "grad_norm": 0.22830356657505035,
      "learning_rate": 0.000776510165648425,
      "loss": 2.1425,
      "step": 729
    },
    {
      "epoch": 0.33470884915176524,
      "grad_norm": 0.2045033872127533,
      "learning_rate": 0.0007758910727230311,
      "loss": 2.0309,
      "step": 730
    },
    {
      "epoch": 0.3351673544245759,
      "grad_norm": 0.4476063549518585,
      "learning_rate": 0.0007752713710790404,
      "loss": 2.0764,
      "step": 731
    },
    {
      "epoch": 0.3356258596973865,
      "grad_norm": 0.22925889492034912,
      "learning_rate": 0.0007746510620837459,
      "loss": 2.1175,
      "step": 732
    },
    {
      "epoch": 0.33608436497019717,
      "grad_norm": 0.21430200338363647,
      "learning_rate": 0.0007740301471057807,
      "loss": 2.0986,
      "step": 733
    },
    {
      "epoch": 0.3365428702430078,
      "grad_norm": 0.21683770418167114,
      "learning_rate": 0.0007734086275151146,
      "loss": 2.1064,
      "step": 734
    },
    {
      "epoch": 0.33700137551581844,
      "grad_norm": 0.20427797734737396,
      "learning_rate": 0.0007727865046830517,
      "loss": 1.9987,
      "step": 735
    },
    {
      "epoch": 0.3374598807886291,
      "grad_norm": 0.20182451605796814,
      "learning_rate": 0.0007721637799822269,
      "loss": 2.0741,
      "step": 736
    },
    {
      "epoch": 0.3379183860614397,
      "grad_norm": 0.22208064794540405,
      "learning_rate": 0.0007715404547866032,
      "loss": 2.0651,
      "step": 737
    },
    {
      "epoch": 0.33837689133425036,
      "grad_norm": 0.2198254019021988,
      "learning_rate": 0.0007709165304714685,
      "loss": 2.1671,
      "step": 738
    },
    {
      "epoch": 0.33883539660706097,
      "grad_norm": 0.21579499542713165,
      "learning_rate": 0.0007702920084134324,
      "loss": 2.1526,
      "step": 739
    },
    {
      "epoch": 0.33929390187987163,
      "grad_norm": 0.2214934527873993,
      "learning_rate": 0.0007696668899904236,
      "loss": 2.0754,
      "step": 740
    },
    {
      "epoch": 0.33975240715268223,
      "grad_norm": 0.2246243953704834,
      "learning_rate": 0.0007690411765816864,
      "loss": 2.1319,
      "step": 741
    },
    {
      "epoch": 0.3402109124254929,
      "grad_norm": 0.2159927785396576,
      "learning_rate": 0.0007684148695677778,
      "loss": 2.0485,
      "step": 742
    },
    {
      "epoch": 0.34066941769830356,
      "grad_norm": 0.21613238751888275,
      "learning_rate": 0.000767787970330565,
      "loss": 2.0544,
      "step": 743
    },
    {
      "epoch": 0.34112792297111416,
      "grad_norm": 0.21271508932113647,
      "learning_rate": 0.000767160480253221,
      "loss": 2.1694,
      "step": 744
    },
    {
      "epoch": 0.3415864282439248,
      "grad_norm": 0.22264498472213745,
      "learning_rate": 0.0007665324007202235,
      "loss": 2.0342,
      "step": 745
    },
    {
      "epoch": 0.3420449335167354,
      "grad_norm": 0.21015164256095886,
      "learning_rate": 0.0007659037331173498,
      "loss": 2.0937,
      "step": 746
    },
    {
      "epoch": 0.3425034387895461,
      "grad_norm": 0.2034616768360138,
      "learning_rate": 0.0007652744788316752,
      "loss": 2.027,
      "step": 747
    },
    {
      "epoch": 0.3429619440623567,
      "grad_norm": 0.2120668739080429,
      "learning_rate": 0.0007646446392515692,
      "loss": 2.0807,
      "step": 748
    },
    {
      "epoch": 0.34342044933516735,
      "grad_norm": 0.21264083683490753,
      "learning_rate": 0.000764014215766693,
      "loss": 2.1034,
      "step": 749
    },
    {
      "epoch": 0.343878954607978,
      "grad_norm": 0.21989795565605164,
      "learning_rate": 0.0007633832097679958,
      "loss": 2.0823,
      "step": 750
    },
    {
      "epoch": 0.3443374598807886,
      "grad_norm": 0.2165154665708542,
      "learning_rate": 0.0007627516226477122,
      "loss": 2.081,
      "step": 751
    },
    {
      "epoch": 0.3447959651535993,
      "grad_norm": 0.2196761816740036,
      "learning_rate": 0.0007621194557993589,
      "loss": 2.1952,
      "step": 752
    },
    {
      "epoch": 0.3452544704264099,
      "grad_norm": 0.21932366490364075,
      "learning_rate": 0.0007614867106177319,
      "loss": 2.0916,
      "step": 753
    },
    {
      "epoch": 0.34571297569922055,
      "grad_norm": 0.2174350470304489,
      "learning_rate": 0.0007608533884989029,
      "loss": 2.1121,
      "step": 754
    },
    {
      "epoch": 0.34617148097203115,
      "grad_norm": 0.20234185457229614,
      "learning_rate": 0.0007602194908402166,
      "loss": 2.0571,
      "step": 755
    },
    {
      "epoch": 0.3466299862448418,
      "grad_norm": 0.20884671807289124,
      "learning_rate": 0.0007595850190402877,
      "loss": 2.0899,
      "step": 756
    },
    {
      "epoch": 0.3470884915176525,
      "grad_norm": 0.2123083919286728,
      "learning_rate": 0.0007589499744989976,
      "loss": 2.2119,
      "step": 757
    },
    {
      "epoch": 0.3475469967904631,
      "grad_norm": 0.21760603785514832,
      "learning_rate": 0.0007583143586174916,
      "loss": 2.1014,
      "step": 758
    },
    {
      "epoch": 0.34800550206327374,
      "grad_norm": 0.20995867252349854,
      "learning_rate": 0.000757678172798175,
      "loss": 2.136,
      "step": 759
    },
    {
      "epoch": 0.34846400733608435,
      "grad_norm": 0.2089584916830063,
      "learning_rate": 0.0007570414184447112,
      "loss": 2.1365,
      "step": 760
    },
    {
      "epoch": 0.348922512608895,
      "grad_norm": 0.21394315361976624,
      "learning_rate": 0.0007564040969620179,
      "loss": 2.1239,
      "step": 761
    },
    {
      "epoch": 0.3493810178817056,
      "grad_norm": 0.21293167769908905,
      "learning_rate": 0.0007557662097562636,
      "loss": 2.009,
      "step": 762
    },
    {
      "epoch": 0.3498395231545163,
      "grad_norm": 0.21346569061279297,
      "learning_rate": 0.0007551277582348658,
      "loss": 2.1954,
      "step": 763
    },
    {
      "epoch": 0.35029802842732694,
      "grad_norm": 0.20872415602207184,
      "learning_rate": 0.0007544887438064862,
      "loss": 2.0825,
      "step": 764
    },
    {
      "epoch": 0.35075653370013754,
      "grad_norm": 0.21596799790859222,
      "learning_rate": 0.0007538491678810294,
      "loss": 2.0125,
      "step": 765
    },
    {
      "epoch": 0.3512150389729482,
      "grad_norm": 0.20811878144741058,
      "learning_rate": 0.0007532090318696381,
      "loss": 2.1042,
      "step": 766
    },
    {
      "epoch": 0.3516735442457588,
      "grad_norm": 0.21239681541919708,
      "learning_rate": 0.0007525683371846913,
      "loss": 2.1107,
      "step": 767
    },
    {
      "epoch": 0.35213204951856947,
      "grad_norm": 0.22132055461406708,
      "learning_rate": 0.0007519270852398001,
      "loss": 2.1116,
      "step": 768
    },
    {
      "epoch": 0.35259055479138013,
      "grad_norm": 0.2060704231262207,
      "learning_rate": 0.000751285277449806,
      "loss": 2.1328,
      "step": 769
    },
    {
      "epoch": 0.35304906006419073,
      "grad_norm": 0.21586249768733978,
      "learning_rate": 0.0007506429152307756,
      "loss": 2.0919,
      "step": 770
    },
    {
      "epoch": 0.3535075653370014,
      "grad_norm": 0.2321099191904068,
      "learning_rate": 0.00075,
      "loss": 2.0824,
      "step": 771
    },
    {
      "epoch": 0.353966070609812,
      "grad_norm": 0.19738395512104034,
      "learning_rate": 0.00074935653317599,
      "loss": 2.075,
      "step": 772
    },
    {
      "epoch": 0.35442457588262266,
      "grad_norm": 0.20850883424282074,
      "learning_rate": 0.000748712516178473,
      "loss": 2.1039,
      "step": 773
    },
    {
      "epoch": 0.35488308115543327,
      "grad_norm": 0.2045629322528839,
      "learning_rate": 0.0007480679504283911,
      "loss": 2.1211,
      "step": 774
    },
    {
      "epoch": 0.3553415864282439,
      "grad_norm": 0.21325182914733887,
      "learning_rate": 0.0007474228373478964,
      "loss": 2.1042,
      "step": 775
    },
    {
      "epoch": 0.3558000917010546,
      "grad_norm": 0.21657821536064148,
      "learning_rate": 0.0007467771783603492,
      "loss": 2.06,
      "step": 776
    },
    {
      "epoch": 0.3562585969738652,
      "grad_norm": 0.20270469784736633,
      "learning_rate": 0.0007461309748903138,
      "loss": 2.1444,
      "step": 777
    },
    {
      "epoch": 0.35671710224667585,
      "grad_norm": 0.2029138207435608,
      "learning_rate": 0.0007454842283635562,
      "loss": 2.0865,
      "step": 778
    },
    {
      "epoch": 0.35717560751948646,
      "grad_norm": 0.20883770287036896,
      "learning_rate": 0.0007448369402070404,
      "loss": 2.0072,
      "step": 779
    },
    {
      "epoch": 0.3576341127922971,
      "grad_norm": 0.22829724848270416,
      "learning_rate": 0.0007441891118489254,
      "loss": 2.1218,
      "step": 780
    },
    {
      "epoch": 0.3580926180651077,
      "grad_norm": 0.20884573459625244,
      "learning_rate": 0.0007435407447185622,
      "loss": 2.1003,
      "step": 781
    },
    {
      "epoch": 0.3585511233379184,
      "grad_norm": 0.21284283697605133,
      "learning_rate": 0.0007428918402464908,
      "loss": 2.0893,
      "step": 782
    },
    {
      "epoch": 0.35900962861072905,
      "grad_norm": 0.20304447412490845,
      "learning_rate": 0.0007422423998644359,
      "loss": 2.0204,
      "step": 783
    },
    {
      "epoch": 0.35946813388353965,
      "grad_norm": 0.2006128877401352,
      "learning_rate": 0.0007415924250053055,
      "loss": 2.0861,
      "step": 784
    },
    {
      "epoch": 0.3599266391563503,
      "grad_norm": 0.22109447419643402,
      "learning_rate": 0.0007409419171031865,
      "loss": 2.0765,
      "step": 785
    },
    {
      "epoch": 0.3603851444291609,
      "grad_norm": 0.21354427933692932,
      "learning_rate": 0.0007402908775933419,
      "loss": 2.0188,
      "step": 786
    },
    {
      "epoch": 0.3608436497019716,
      "grad_norm": 0.22207674384117126,
      "learning_rate": 0.0007396393079122077,
      "loss": 1.9912,
      "step": 787
    },
    {
      "epoch": 0.3613021549747822,
      "grad_norm": 0.21520085632801056,
      "learning_rate": 0.0007389872094973896,
      "loss": 2.025,
      "step": 788
    },
    {
      "epoch": 0.36176066024759285,
      "grad_norm": 0.2103038728237152,
      "learning_rate": 0.00073833458378766,
      "loss": 2.0711,
      "step": 789
    },
    {
      "epoch": 0.3622191655204035,
      "grad_norm": 0.22774966061115265,
      "learning_rate": 0.0007376814322229544,
      "loss": 2.1701,
      "step": 790
    },
    {
      "epoch": 0.3626776707932141,
      "grad_norm": 0.21860061585903168,
      "learning_rate": 0.0007370277562443688,
      "loss": 2.0727,
      "step": 791
    },
    {
      "epoch": 0.3631361760660248,
      "grad_norm": 0.21404291689395905,
      "learning_rate": 0.0007363735572941564,
      "loss": 2.1295,
      "step": 792
    },
    {
      "epoch": 0.3635946813388354,
      "grad_norm": 0.23994508385658264,
      "learning_rate": 0.0007357188368157236,
      "loss": 2.2336,
      "step": 793
    },
    {
      "epoch": 0.36405318661164604,
      "grad_norm": 0.22355128824710846,
      "learning_rate": 0.0007350635962536284,
      "loss": 2.0792,
      "step": 794
    },
    {
      "epoch": 0.36451169188445665,
      "grad_norm": 0.21207444369792938,
      "learning_rate": 0.0007344078370535756,
      "loss": 2.0298,
      "step": 795
    },
    {
      "epoch": 0.3649701971572673,
      "grad_norm": 0.19830991327762604,
      "learning_rate": 0.0007337515606624148,
      "loss": 2.0571,
      "step": 796
    },
    {
      "epoch": 0.36542870243007797,
      "grad_norm": 0.20958855748176575,
      "learning_rate": 0.0007330947685281362,
      "loss": 2.0235,
      "step": 797
    },
    {
      "epoch": 0.3658872077028886,
      "grad_norm": 0.2166910022497177,
      "learning_rate": 0.0007324374620998682,
      "loss": 2.0901,
      "step": 798
    },
    {
      "epoch": 0.36634571297569923,
      "grad_norm": 0.22643771767616272,
      "learning_rate": 0.000731779642827874,
      "loss": 2.0223,
      "step": 799
    },
    {
      "epoch": 0.36680421824850984,
      "grad_norm": 0.22884224355220795,
      "learning_rate": 0.0007311213121635483,
      "loss": 2.0692,
      "step": 800
    },
    {
      "epoch": 0.3672627235213205,
      "grad_norm": 0.23521332442760468,
      "learning_rate": 0.0007304624715594139,
      "loss": 2.0613,
      "step": 801
    },
    {
      "epoch": 0.3677212287941311,
      "grad_norm": 0.22290343046188354,
      "learning_rate": 0.0007298031224691193,
      "loss": 2.1168,
      "step": 802
    },
    {
      "epoch": 0.36817973406694177,
      "grad_norm": 0.22263464331626892,
      "learning_rate": 0.0007291432663474339,
      "loss": 2.0489,
      "step": 803
    },
    {
      "epoch": 0.3686382393397524,
      "grad_norm": 0.2173481583595276,
      "learning_rate": 0.0007284829046502467,
      "loss": 2.0596,
      "step": 804
    },
    {
      "epoch": 0.36909674461256303,
      "grad_norm": 0.20734253525733948,
      "learning_rate": 0.0007278220388345619,
      "loss": 2.0702,
      "step": 805
    },
    {
      "epoch": 0.3695552498853737,
      "grad_norm": 0.20500902831554413,
      "learning_rate": 0.0007271606703584958,
      "loss": 2.0342,
      "step": 806
    },
    {
      "epoch": 0.3700137551581843,
      "grad_norm": 0.20836085081100464,
      "learning_rate": 0.000726498800681274,
      "loss": 2.0605,
      "step": 807
    },
    {
      "epoch": 0.37047226043099496,
      "grad_norm": 0.2043185532093048,
      "learning_rate": 0.0007258364312632279,
      "loss": 2.1001,
      "step": 808
    },
    {
      "epoch": 0.3709307657038056,
      "grad_norm": 0.20976318418979645,
      "learning_rate": 0.0007251735635657915,
      "loss": 2.0399,
      "step": 809
    },
    {
      "epoch": 0.3713892709766162,
      "grad_norm": 0.22801978886127472,
      "learning_rate": 0.000724510199051498,
      "loss": 2.0948,
      "step": 810
    },
    {
      "epoch": 0.3718477762494269,
      "grad_norm": 0.22062665224075317,
      "learning_rate": 0.0007238463391839769,
      "loss": 2.0827,
      "step": 811
    },
    {
      "epoch": 0.3723062815222375,
      "grad_norm": 0.21854837238788605,
      "learning_rate": 0.0007231819854279508,
      "loss": 2.1484,
      "step": 812
    },
    {
      "epoch": 0.37276478679504815,
      "grad_norm": 0.20922479033470154,
      "learning_rate": 0.0007225171392492316,
      "loss": 2.0454,
      "step": 813
    },
    {
      "epoch": 0.37322329206785876,
      "grad_norm": 0.22007781267166138,
      "learning_rate": 0.0007218518021147182,
      "loss": 2.1219,
      "step": 814
    },
    {
      "epoch": 0.3736817973406694,
      "grad_norm": 0.2107928842306137,
      "learning_rate": 0.0007211859754923923,
      "loss": 2.0853,
      "step": 815
    },
    {
      "epoch": 0.3741403026134801,
      "grad_norm": 0.21697530150413513,
      "learning_rate": 0.0007205196608513158,
      "loss": 2.0384,
      "step": 816
    },
    {
      "epoch": 0.3745988078862907,
      "grad_norm": 0.2039753496646881,
      "learning_rate": 0.0007198528596616272,
      "loss": 2.1486,
      "step": 817
    },
    {
      "epoch": 0.37505731315910135,
      "grad_norm": 0.20646320283412933,
      "learning_rate": 0.0007191855733945387,
      "loss": 1.9986,
      "step": 818
    },
    {
      "epoch": 0.37551581843191195,
      "grad_norm": 0.19979228079319,
      "learning_rate": 0.0007185178035223327,
      "loss": 1.9999,
      "step": 819
    },
    {
      "epoch": 0.3759743237047226,
      "grad_norm": 0.217938631772995,
      "learning_rate": 0.0007178495515183583,
      "loss": 2.0603,
      "step": 820
    },
    {
      "epoch": 0.3764328289775332,
      "grad_norm": 0.21170467138290405,
      "learning_rate": 0.000717180818857029,
      "loss": 2.0751,
      "step": 821
    },
    {
      "epoch": 0.3768913342503439,
      "grad_norm": 0.2078528255224228,
      "learning_rate": 0.0007165116070138182,
      "loss": 1.9768,
      "step": 822
    },
    {
      "epoch": 0.37734983952315454,
      "grad_norm": 0.2027733325958252,
      "learning_rate": 0.0007158419174652569,
      "loss": 1.9606,
      "step": 823
    },
    {
      "epoch": 0.37780834479596515,
      "grad_norm": 0.2114807814359665,
      "learning_rate": 0.00071517175168893,
      "loss": 2.0479,
      "step": 824
    },
    {
      "epoch": 0.3782668500687758,
      "grad_norm": 0.20815566182136536,
      "learning_rate": 0.0007145011111634732,
      "loss": 2.1592,
      "step": 825
    },
    {
      "epoch": 0.3787253553415864,
      "grad_norm": 0.21640852093696594,
      "learning_rate": 0.0007138299973685694,
      "loss": 2.1328,
      "step": 826
    },
    {
      "epoch": 0.3791838606143971,
      "grad_norm": 0.21865004301071167,
      "learning_rate": 0.0007131584117849459,
      "loss": 2.07,
      "step": 827
    },
    {
      "epoch": 0.3796423658872077,
      "grad_norm": 0.2109341025352478,
      "learning_rate": 0.0007124863558943713,
      "loss": 2.0998,
      "step": 828
    },
    {
      "epoch": 0.38010087116001834,
      "grad_norm": 0.2075556218624115,
      "learning_rate": 0.0007118138311796514,
      "loss": 2.1325,
      "step": 829
    },
    {
      "epoch": 0.380559376432829,
      "grad_norm": 0.21273428201675415,
      "learning_rate": 0.0007111408391246262,
      "loss": 1.9966,
      "step": 830
    },
    {
      "epoch": 0.3810178817056396,
      "grad_norm": 0.21335318684577942,
      "learning_rate": 0.0007104673812141675,
      "loss": 2.087,
      "step": 831
    },
    {
      "epoch": 0.38147638697845027,
      "grad_norm": 0.2303408533334732,
      "learning_rate": 0.0007097934589341745,
      "loss": 2.1438,
      "step": 832
    },
    {
      "epoch": 0.38193489225126087,
      "grad_norm": 0.20290611684322357,
      "learning_rate": 0.0007091190737715711,
      "loss": 2.1558,
      "step": 833
    },
    {
      "epoch": 0.38239339752407153,
      "grad_norm": 0.20176394283771515,
      "learning_rate": 0.0007084442272143026,
      "loss": 2.1098,
      "step": 834
    },
    {
      "epoch": 0.38285190279688214,
      "grad_norm": 0.20101198554039001,
      "learning_rate": 0.000707768920751332,
      "loss": 2.1348,
      "step": 835
    },
    {
      "epoch": 0.3833104080696928,
      "grad_norm": 0.20813648402690887,
      "learning_rate": 0.0007070931558726373,
      "loss": 2.1458,
      "step": 836
    },
    {
      "epoch": 0.38376891334250346,
      "grad_norm": 0.20258378982543945,
      "learning_rate": 0.0007064169340692076,
      "loss": 2.0493,
      "step": 837
    },
    {
      "epoch": 0.38422741861531406,
      "grad_norm": 0.22148089110851288,
      "learning_rate": 0.0007057402568330407,
      "loss": 1.9936,
      "step": 838
    },
    {
      "epoch": 0.3846859238881247,
      "grad_norm": 0.22553066909313202,
      "learning_rate": 0.0007050631256571389,
      "loss": 2.0529,
      "step": 839
    },
    {
      "epoch": 0.38514442916093533,
      "grad_norm": 0.22510316967964172,
      "learning_rate": 0.000704385542035506,
      "loss": 2.0702,
      "step": 840
    },
    {
      "epoch": 0.385602934433746,
      "grad_norm": 0.21246202290058136,
      "learning_rate": 0.000703707507463144,
      "loss": 2.0031,
      "step": 841
    },
    {
      "epoch": 0.38606143970655665,
      "grad_norm": 0.2128654569387436,
      "learning_rate": 0.0007030290234360505,
      "loss": 2.1782,
      "step": 842
    },
    {
      "epoch": 0.38651994497936726,
      "grad_norm": 0.2170751690864563,
      "learning_rate": 0.0007023500914512139,
      "loss": 2.0904,
      "step": 843
    },
    {
      "epoch": 0.3869784502521779,
      "grad_norm": 0.20245559513568878,
      "learning_rate": 0.0007016707130066116,
      "loss": 2.0406,
      "step": 844
    },
    {
      "epoch": 0.3874369555249885,
      "grad_norm": 0.2047078013420105,
      "learning_rate": 0.0007009908896012055,
      "loss": 2.0758,
      "step": 845
    },
    {
      "epoch": 0.3878954607977992,
      "grad_norm": 0.19959738850593567,
      "learning_rate": 0.0007003106227349399,
      "loss": 2.1395,
      "step": 846
    },
    {
      "epoch": 0.3883539660706098,
      "grad_norm": 0.2197038233280182,
      "learning_rate": 0.000699629913908737,
      "loss": 2.1151,
      "step": 847
    },
    {
      "epoch": 0.38881247134342045,
      "grad_norm": 0.2054535448551178,
      "learning_rate": 0.0006989487646244943,
      "loss": 2.087,
      "step": 848
    },
    {
      "epoch": 0.3892709766162311,
      "grad_norm": 0.22070245444774628,
      "learning_rate": 0.0006982671763850814,
      "loss": 2.0573,
      "step": 849
    },
    {
      "epoch": 0.3897294818890417,
      "grad_norm": 0.21630869805812836,
      "learning_rate": 0.0006975851506943359,
      "loss": 2.0464,
      "step": 850
    },
    {
      "epoch": 0.3901879871618524,
      "grad_norm": 0.21979336440563202,
      "learning_rate": 0.0006969026890570611,
      "loss": 2.1261,
      "step": 851
    },
    {
      "epoch": 0.390646492434663,
      "grad_norm": 0.20932531356811523,
      "learning_rate": 0.0006962197929790216,
      "loss": 1.9993,
      "step": 852
    },
    {
      "epoch": 0.39110499770747365,
      "grad_norm": 0.2068345546722412,
      "learning_rate": 0.0006955364639669409,
      "loss": 2.136,
      "step": 853
    },
    {
      "epoch": 0.39156350298028425,
      "grad_norm": 0.21895228326320648,
      "learning_rate": 0.0006948527035284978,
      "loss": 2.0905,
      "step": 854
    },
    {
      "epoch": 0.3920220082530949,
      "grad_norm": 0.21093374490737915,
      "learning_rate": 0.0006941685131723225,
      "loss": 2.1524,
      "step": 855
    },
    {
      "epoch": 0.39248051352590557,
      "grad_norm": 0.2053462713956833,
      "learning_rate": 0.0006934838944079943,
      "loss": 2.0794,
      "step": 856
    },
    {
      "epoch": 0.3929390187987162,
      "grad_norm": 0.2091938555240631,
      "learning_rate": 0.0006927988487460378,
      "loss": 2.0638,
      "step": 857
    },
    {
      "epoch": 0.39339752407152684,
      "grad_norm": 0.21093951165676117,
      "learning_rate": 0.0006921133776979186,
      "loss": 2.0885,
      "step": 858
    },
    {
      "epoch": 0.39385602934433744,
      "grad_norm": 0.21952418982982635,
      "learning_rate": 0.0006914274827760418,
      "loss": 2.05,
      "step": 859
    },
    {
      "epoch": 0.3943145346171481,
      "grad_norm": 0.20889170467853546,
      "learning_rate": 0.0006907411654937475,
      "loss": 2.1349,
      "step": 860
    },
    {
      "epoch": 0.3947730398899587,
      "grad_norm": 0.22418485581874847,
      "learning_rate": 0.0006900544273653075,
      "loss": 2.1942,
      "step": 861
    },
    {
      "epoch": 0.39523154516276937,
      "grad_norm": 0.21914507448673248,
      "learning_rate": 0.000689367269905922,
      "loss": 2.1118,
      "step": 862
    },
    {
      "epoch": 0.39569005043558003,
      "grad_norm": 0.2189093828201294,
      "learning_rate": 0.0006886796946317168,
      "loss": 2.0972,
      "step": 863
    },
    {
      "epoch": 0.39614855570839064,
      "grad_norm": 0.19997408986091614,
      "learning_rate": 0.0006879917030597397,
      "loss": 2.0568,
      "step": 864
    },
    {
      "epoch": 0.3966070609812013,
      "grad_norm": 0.2242313027381897,
      "learning_rate": 0.0006873032967079561,
      "loss": 2.0505,
      "step": 865
    },
    {
      "epoch": 0.3970655662540119,
      "grad_norm": 0.2161043882369995,
      "learning_rate": 0.0006866144770952474,
      "loss": 2.0573,
      "step": 866
    },
    {
      "epoch": 0.39752407152682256,
      "grad_norm": 0.2095109522342682,
      "learning_rate": 0.0006859252457414067,
      "loss": 2.1499,
      "step": 867
    },
    {
      "epoch": 0.39798257679963317,
      "grad_norm": 0.231190025806427,
      "learning_rate": 0.0006852356041671351,
      "loss": 2.0066,
      "step": 868
    },
    {
      "epoch": 0.39844108207244383,
      "grad_norm": 0.2327369600534439,
      "learning_rate": 0.0006845455538940394,
      "loss": 2.1089,
      "step": 869
    },
    {
      "epoch": 0.3988995873452545,
      "grad_norm": 0.20355793833732605,
      "learning_rate": 0.0006838550964446276,
      "loss": 2.0468,
      "step": 870
    },
    {
      "epoch": 0.3993580926180651,
      "grad_norm": 0.22456923127174377,
      "learning_rate": 0.0006831642333423067,
      "loss": 2.1988,
      "step": 871
    },
    {
      "epoch": 0.39981659789087576,
      "grad_norm": 0.22627700865268707,
      "learning_rate": 0.000682472966111378,
      "loss": 2.0785,
      "step": 872
    },
    {
      "epoch": 0.40027510316368636,
      "grad_norm": 0.2160479724407196,
      "learning_rate": 0.0006817812962770348,
      "loss": 2.0925,
      "step": 873
    },
    {
      "epoch": 0.400733608436497,
      "grad_norm": 0.2075222283601761,
      "learning_rate": 0.0006810892253653589,
      "loss": 2.0357,
      "step": 874
    },
    {
      "epoch": 0.40119211370930763,
      "grad_norm": 0.2117258459329605,
      "learning_rate": 0.0006803967549033167,
      "loss": 2.0846,
      "step": 875
    },
    {
      "epoch": 0.4016506189821183,
      "grad_norm": 0.23714156448841095,
      "learning_rate": 0.0006797038864187564,
      "loss": 2.1309,
      "step": 876
    },
    {
      "epoch": 0.40210912425492895,
      "grad_norm": 0.22934666275978088,
      "learning_rate": 0.0006790106214404043,
      "loss": 2.1396,
      "step": 877
    },
    {
      "epoch": 0.40256762952773956,
      "grad_norm": 0.19950629770755768,
      "learning_rate": 0.0006783169614978614,
      "loss": 2.0773,
      "step": 878
    },
    {
      "epoch": 0.4030261348005502,
      "grad_norm": 0.20867161452770233,
      "learning_rate": 0.0006776229081216001,
      "loss": 2.0962,
      "step": 879
    },
    {
      "epoch": 0.4034846400733608,
      "grad_norm": 0.21709847450256348,
      "learning_rate": 0.0006769284628429611,
      "loss": 2.0256,
      "step": 880
    },
    {
      "epoch": 0.4039431453461715,
      "grad_norm": 0.21099695563316345,
      "learning_rate": 0.0006762336271941498,
      "loss": 2.0795,
      "step": 881
    },
    {
      "epoch": 0.40440165061898214,
      "grad_norm": 0.20992910861968994,
      "learning_rate": 0.0006755384027082326,
      "loss": 2.0694,
      "step": 882
    },
    {
      "epoch": 0.40486015589179275,
      "grad_norm": 0.2164323627948761,
      "learning_rate": 0.0006748427909191342,
      "loss": 2.0251,
      "step": 883
    },
    {
      "epoch": 0.4053186611646034,
      "grad_norm": 0.2231670469045639,
      "learning_rate": 0.0006741467933616335,
      "loss": 2.0882,
      "step": 884
    },
    {
      "epoch": 0.405777166437414,
      "grad_norm": 0.21093811094760895,
      "learning_rate": 0.0006734504115713604,
      "loss": 2.03,
      "step": 885
    },
    {
      "epoch": 0.4062356717102247,
      "grad_norm": 0.21806208789348602,
      "learning_rate": 0.0006727536470847932,
      "loss": 2.0615,
      "step": 886
    },
    {
      "epoch": 0.4066941769830353,
      "grad_norm": 0.22733865678310394,
      "learning_rate": 0.000672056501439254,
      "loss": 2.0364,
      "step": 887
    },
    {
      "epoch": 0.40715268225584594,
      "grad_norm": 0.23784035444259644,
      "learning_rate": 0.0006713589761729063,
      "loss": 2.0437,
      "step": 888
    },
    {
      "epoch": 0.4076111875286566,
      "grad_norm": 0.2130998969078064,
      "learning_rate": 0.0006706610728247508,
      "loss": 2.1002,
      "step": 889
    },
    {
      "epoch": 0.4080696928014672,
      "grad_norm": 0.20843441784381866,
      "learning_rate": 0.0006699627929346227,
      "loss": 2.1252,
      "step": 890
    },
    {
      "epoch": 0.40852819807427787,
      "grad_norm": 0.21146291494369507,
      "learning_rate": 0.0006692641380431879,
      "loss": 2.0309,
      "step": 891
    },
    {
      "epoch": 0.4089867033470885,
      "grad_norm": 0.20947138965129852,
      "learning_rate": 0.0006685651096919393,
      "loss": 2.0659,
      "step": 892
    },
    {
      "epoch": 0.40944520861989914,
      "grad_norm": 0.22898398339748383,
      "learning_rate": 0.0006678657094231944,
      "loss": 1.9026,
      "step": 893
    },
    {
      "epoch": 0.40990371389270974,
      "grad_norm": 0.21516677737236023,
      "learning_rate": 0.0006671659387800909,
      "loss": 2.2066,
      "step": 894
    },
    {
      "epoch": 0.4103622191655204,
      "grad_norm": 0.20841491222381592,
      "learning_rate": 0.000666465799306584,
      "loss": 2.0552,
      "step": 895
    },
    {
      "epoch": 0.41082072443833106,
      "grad_norm": 0.21478383243083954,
      "learning_rate": 0.0006657652925474423,
      "loss": 2.1004,
      "step": 896
    },
    {
      "epoch": 0.41127922971114167,
      "grad_norm": 0.2086239606142044,
      "learning_rate": 0.000665064420048245,
      "loss": 2.1073,
      "step": 897
    },
    {
      "epoch": 0.41173773498395233,
      "grad_norm": 0.2139568030834198,
      "learning_rate": 0.0006643631833553785,
      "loss": 2.0216,
      "step": 898
    },
    {
      "epoch": 0.41219624025676294,
      "grad_norm": 0.20360791683197021,
      "learning_rate": 0.000663661584016032,
      "loss": 2.0235,
      "step": 899
    },
    {
      "epoch": 0.4126547455295736,
      "grad_norm": 0.21482908725738525,
      "learning_rate": 0.0006629596235781957,
      "loss": 2.1346,
      "step": 900
    },
    {
      "epoch": 0.4131132508023842,
      "grad_norm": 0.21680350601673126,
      "learning_rate": 0.0006622573035906556,
      "loss": 2.0778,
      "step": 901
    },
    {
      "epoch": 0.41357175607519486,
      "grad_norm": 0.21091042459011078,
      "learning_rate": 0.0006615546256029921,
      "loss": 2.1193,
      "step": 902
    },
    {
      "epoch": 0.4140302613480055,
      "grad_norm": 0.21335673332214355,
      "learning_rate": 0.0006608515911655743,
      "loss": 2.2338,
      "step": 903
    },
    {
      "epoch": 0.41448876662081613,
      "grad_norm": 0.21011531352996826,
      "learning_rate": 0.0006601482018295591,
      "loss": 2.1079,
      "step": 904
    },
    {
      "epoch": 0.4149472718936268,
      "grad_norm": 0.20255854725837708,
      "learning_rate": 0.0006594444591468851,
      "loss": 2.1266,
      "step": 905
    },
    {
      "epoch": 0.4154057771664374,
      "grad_norm": 0.20041951537132263,
      "learning_rate": 0.0006587403646702713,
      "loss": 2.1118,
      "step": 906
    },
    {
      "epoch": 0.41586428243924806,
      "grad_norm": 0.21109986305236816,
      "learning_rate": 0.0006580359199532126,
      "loss": 2.0559,
      "step": 907
    },
    {
      "epoch": 0.41632278771205866,
      "grad_norm": 0.215422123670578,
      "learning_rate": 0.000657331126549977,
      "loss": 2.018,
      "step": 908
    },
    {
      "epoch": 0.4167812929848693,
      "grad_norm": 0.19280295073986053,
      "learning_rate": 0.0006566259860156014,
      "loss": 2.0348,
      "step": 909
    },
    {
      "epoch": 0.41723979825768,
      "grad_norm": 0.2146463245153427,
      "learning_rate": 0.0006559204999058888,
      "loss": 2.0782,
      "step": 910
    },
    {
      "epoch": 0.4176983035304906,
      "grad_norm": 0.22356821596622467,
      "learning_rate": 0.0006552146697774049,
      "loss": 2.1067,
      "step": 911
    },
    {
      "epoch": 0.41815680880330125,
      "grad_norm": 0.2285921573638916,
      "learning_rate": 0.0006545084971874737,
      "loss": 2.089,
      "step": 912
    },
    {
      "epoch": 0.41861531407611186,
      "grad_norm": 0.21373477578163147,
      "learning_rate": 0.0006538019836941758,
      "loss": 2.1091,
      "step": 913
    },
    {
      "epoch": 0.4190738193489225,
      "grad_norm": 0.2222813069820404,
      "learning_rate": 0.0006530951308563431,
      "loss": 2.1372,
      "step": 914
    },
    {
      "epoch": 0.4195323246217332,
      "grad_norm": 0.20782190561294556,
      "learning_rate": 0.0006523879402335567,
      "loss": 2.0772,
      "step": 915
    },
    {
      "epoch": 0.4199908298945438,
      "grad_norm": 0.20705938339233398,
      "learning_rate": 0.0006516804133861429,
      "loss": 2.0522,
      "step": 916
    },
    {
      "epoch": 0.42044933516735444,
      "grad_norm": 0.1976783722639084,
      "learning_rate": 0.0006509725518751698,
      "loss": 2.0227,
      "step": 917
    },
    {
      "epoch": 0.42090784044016505,
      "grad_norm": 0.21324476599693298,
      "learning_rate": 0.0006502643572624438,
      "loss": 2.0942,
      "step": 918
    },
    {
      "epoch": 0.4213663457129757,
      "grad_norm": 0.21022439002990723,
      "learning_rate": 0.0006495558311105064,
      "loss": 2.0578,
      "step": 919
    },
    {
      "epoch": 0.4218248509857863,
      "grad_norm": 0.19856728613376617,
      "learning_rate": 0.0006488469749826305,
      "loss": 2.0783,
      "step": 920
    },
    {
      "epoch": 0.422283356258597,
      "grad_norm": 0.2112470418214798,
      "learning_rate": 0.000648137790442817,
      "loss": 2.1032,
      "step": 921
    },
    {
      "epoch": 0.42274186153140764,
      "grad_norm": 0.19788049161434174,
      "learning_rate": 0.0006474282790557916,
      "loss": 2.0256,
      "step": 922
    },
    {
      "epoch": 0.42320036680421824,
      "grad_norm": 0.21017703413963318,
      "learning_rate": 0.000646718442387001,
      "loss": 2.0573,
      "step": 923
    },
    {
      "epoch": 0.4236588720770289,
      "grad_norm": 0.19944459199905396,
      "learning_rate": 0.0006460082820026094,
      "loss": 2.0308,
      "step": 924
    },
    {
      "epoch": 0.4241173773498395,
      "grad_norm": 0.19836074113845825,
      "learning_rate": 0.0006452977994694959,
      "loss": 1.9801,
      "step": 925
    },
    {
      "epoch": 0.42457588262265017,
      "grad_norm": 0.21248267590999603,
      "learning_rate": 0.0006445869963552496,
      "loss": 2.0362,
      "step": 926
    },
    {
      "epoch": 0.4250343878954608,
      "grad_norm": 0.20802991092205048,
      "learning_rate": 0.0006438758742281672,
      "loss": 2.067,
      "step": 927
    },
    {
      "epoch": 0.42549289316827144,
      "grad_norm": 0.19991548359394073,
      "learning_rate": 0.0006431644346572495,
      "loss": 2.0502,
      "step": 928
    },
    {
      "epoch": 0.4259513984410821,
      "grad_norm": 0.21240085363388062,
      "learning_rate": 0.0006424526792121974,
      "loss": 1.987,
      "step": 929
    },
    {
      "epoch": 0.4264099037138927,
      "grad_norm": 0.20751000940799713,
      "learning_rate": 0.0006417406094634089,
      "loss": 2.1249,
      "step": 930
    },
    {
      "epoch": 0.42686840898670336,
      "grad_norm": 0.2003788948059082,
      "learning_rate": 0.0006410282269819756,
      "loss": 2.0471,
      "step": 931
    },
    {
      "epoch": 0.42732691425951397,
      "grad_norm": 0.2073497325181961,
      "learning_rate": 0.0006403155333396787,
      "loss": 2.1344,
      "step": 932
    },
    {
      "epoch": 0.42778541953232463,
      "grad_norm": 0.20076197385787964,
      "learning_rate": 0.0006396025301089863,
      "loss": 2.1099,
      "step": 933
    },
    {
      "epoch": 0.42824392480513523,
      "grad_norm": 0.20571136474609375,
      "learning_rate": 0.0006388892188630493,
      "loss": 2.027,
      "step": 934
    },
    {
      "epoch": 0.4287024300779459,
      "grad_norm": 0.20998746156692505,
      "learning_rate": 0.0006381756011756982,
      "loss": 2.088,
      "step": 935
    },
    {
      "epoch": 0.42916093535075656,
      "grad_norm": 0.2042720764875412,
      "learning_rate": 0.0006374616786214403,
      "loss": 2.1478,
      "step": 936
    },
    {
      "epoch": 0.42961944062356716,
      "grad_norm": 0.19962932169437408,
      "learning_rate": 0.0006367474527754544,
      "loss": 2.0725,
      "step": 937
    },
    {
      "epoch": 0.4300779458963778,
      "grad_norm": 0.19475029408931732,
      "learning_rate": 0.0006360329252135894,
      "loss": 2.1002,
      "step": 938
    },
    {
      "epoch": 0.4305364511691884,
      "grad_norm": 0.20193517208099365,
      "learning_rate": 0.0006353180975123595,
      "loss": 2.1056,
      "step": 939
    },
    {
      "epoch": 0.4309949564419991,
      "grad_norm": 0.20809857547283173,
      "learning_rate": 0.0006346029712489413,
      "loss": 2.0884,
      "step": 940
    },
    {
      "epoch": 0.4314534617148097,
      "grad_norm": 0.20230801403522491,
      "learning_rate": 0.0006338875480011698,
      "loss": 2.1704,
      "step": 941
    },
    {
      "epoch": 0.43191196698762035,
      "grad_norm": 0.21519893407821655,
      "learning_rate": 0.0006331718293475357,
      "loss": 1.9686,
      "step": 942
    },
    {
      "epoch": 0.432370472260431,
      "grad_norm": 0.21167972683906555,
      "learning_rate": 0.0006324558168671811,
      "loss": 2.0717,
      "step": 943
    },
    {
      "epoch": 0.4328289775332416,
      "grad_norm": 0.21352535486221313,
      "learning_rate": 0.0006317395121398968,
      "loss": 2.1202,
      "step": 944
    },
    {
      "epoch": 0.4332874828060523,
      "grad_norm": 0.21599027514457703,
      "learning_rate": 0.0006310229167461179,
      "loss": 2.0645,
      "step": 945
    },
    {
      "epoch": 0.4337459880788629,
      "grad_norm": 0.21887485682964325,
      "learning_rate": 0.0006303060322669214,
      "loss": 1.9267,
      "step": 946
    },
    {
      "epoch": 0.43420449335167355,
      "grad_norm": 0.2302350401878357,
      "learning_rate": 0.0006295888602840214,
      "loss": 2.1943,
      "step": 947
    },
    {
      "epoch": 0.4346629986244842,
      "grad_norm": 0.22882460057735443,
      "learning_rate": 0.0006288714023797671,
      "loss": 2.063,
      "step": 948
    },
    {
      "epoch": 0.4351215038972948,
      "grad_norm": 0.22170476615428925,
      "learning_rate": 0.000628153660137138,
      "loss": 2.1885,
      "step": 949
    },
    {
      "epoch": 0.4355800091701055,
      "grad_norm": 0.22310447692871094,
      "learning_rate": 0.0006274356351397413,
      "loss": 2.0855,
      "step": 950
    },
    {
      "epoch": 0.4360385144429161,
      "grad_norm": 0.20974768698215485,
      "learning_rate": 0.0006267173289718079,
      "loss": 2.0232,
      "step": 951
    },
    {
      "epoch": 0.43649701971572674,
      "grad_norm": 0.21021468937397003,
      "learning_rate": 0.000625998743218189,
      "loss": 2.0996,
      "step": 952
    },
    {
      "epoch": 0.43695552498853735,
      "grad_norm": 0.20859314501285553,
      "learning_rate": 0.000625279879464353,
      "loss": 2.0254,
      "step": 953
    },
    {
      "epoch": 0.437414030261348,
      "grad_norm": 0.21989208459854126,
      "learning_rate": 0.000624560739296381,
      "loss": 2.0756,
      "step": 954
    },
    {
      "epoch": 0.43787253553415867,
      "grad_norm": 0.20779722929000854,
      "learning_rate": 0.0006238413243009648,
      "loss": 2.0149,
      "step": 955
    },
    {
      "epoch": 0.4383310408069693,
      "grad_norm": 0.2116817831993103,
      "learning_rate": 0.000623121636065402,
      "loss": 2.0841,
      "step": 956
    },
    {
      "epoch": 0.43878954607977994,
      "grad_norm": 0.21340878307819366,
      "learning_rate": 0.0006224016761775933,
      "loss": 2.0406,
      "step": 957
    },
    {
      "epoch": 0.43924805135259054,
      "grad_norm": 0.2265952080488205,
      "learning_rate": 0.0006216814462260386,
      "loss": 2.1389,
      "step": 958
    },
    {
      "epoch": 0.4397065566254012,
      "grad_norm": 0.20896878838539124,
      "learning_rate": 0.0006209609477998338,
      "loss": 2.007,
      "step": 959
    },
    {
      "epoch": 0.4401650618982118,
      "grad_norm": 0.2125600427389145,
      "learning_rate": 0.0006202401824886674,
      "loss": 2.1075,
      "step": 960
    },
    {
      "epoch": 0.44062356717102247,
      "grad_norm": 0.22515566647052765,
      "learning_rate": 0.0006195191518828162,
      "loss": 2.0821,
      "step": 961
    },
    {
      "epoch": 0.44108207244383313,
      "grad_norm": 0.20651713013648987,
      "learning_rate": 0.0006187978575731427,
      "loss": 2.0584,
      "step": 962
    },
    {
      "epoch": 0.44154057771664373,
      "grad_norm": 0.21853046119213104,
      "learning_rate": 0.0006180763011510911,
      "loss": 2.1092,
      "step": 963
    },
    {
      "epoch": 0.4419990829894544,
      "grad_norm": 0.21378421783447266,
      "learning_rate": 0.000617354484208684,
      "loss": 2.0178,
      "step": 964
    },
    {
      "epoch": 0.442457588262265,
      "grad_norm": 0.20787006616592407,
      "learning_rate": 0.0006166324083385189,
      "loss": 2.0704,
      "step": 965
    },
    {
      "epoch": 0.44291609353507566,
      "grad_norm": 0.21094202995300293,
      "learning_rate": 0.0006159100751337642,
      "loss": 2.1089,
      "step": 966
    },
    {
      "epoch": 0.44337459880788627,
      "grad_norm": 0.20137406885623932,
      "learning_rate": 0.0006151874861881565,
      "loss": 2.0933,
      "step": 967
    },
    {
      "epoch": 0.4438331040806969,
      "grad_norm": 0.2027188241481781,
      "learning_rate": 0.0006144646430959964,
      "loss": 2.0202,
      "step": 968
    },
    {
      "epoch": 0.4442916093535076,
      "grad_norm": 0.20864196121692657,
      "learning_rate": 0.0006137415474521454,
      "loss": 2.0501,
      "step": 969
    },
    {
      "epoch": 0.4447501146263182,
      "grad_norm": 0.2138124406337738,
      "learning_rate": 0.0006130182008520222,
      "loss": 2.0641,
      "step": 970
    },
    {
      "epoch": 0.44520861989912885,
      "grad_norm": 0.20460017025470734,
      "learning_rate": 0.000612294604891599,
      "loss": 2.07,
      "step": 971
    },
    {
      "epoch": 0.44566712517193946,
      "grad_norm": 0.21167565882205963,
      "learning_rate": 0.0006115707611673986,
      "loss": 2.085,
      "step": 972
    },
    {
      "epoch": 0.4461256304447501,
      "grad_norm": 0.2213941067457199,
      "learning_rate": 0.0006108466712764902,
      "loss": 2.1852,
      "step": 973
    },
    {
      "epoch": 0.4465841357175607,
      "grad_norm": 0.2010621726512909,
      "learning_rate": 0.0006101223368164858,
      "loss": 2.0341,
      "step": 974
    },
    {
      "epoch": 0.4470426409903714,
      "grad_norm": 0.20589584112167358,
      "learning_rate": 0.0006093977593855375,
      "loss": 2.1516,
      "step": 975
    },
    {
      "epoch": 0.44750114626318205,
      "grad_norm": 0.2091546356678009,
      "learning_rate": 0.0006086729405823335,
      "loss": 2.0272,
      "step": 976
    },
    {
      "epoch": 0.44795965153599265,
      "grad_norm": 0.217277392745018,
      "learning_rate": 0.0006079478820060943,
      "loss": 2.0718,
      "step": 977
    },
    {
      "epoch": 0.4484181568088033,
      "grad_norm": 0.1900898963212967,
      "learning_rate": 0.0006072225852565695,
      "loss": 2.0771,
      "step": 978
    },
    {
      "epoch": 0.4488766620816139,
      "grad_norm": 0.20387324690818787,
      "learning_rate": 0.0006064970519340341,
      "loss": 2.0427,
      "step": 979
    },
    {
      "epoch": 0.4493351673544246,
      "grad_norm": 0.20544226467609406,
      "learning_rate": 0.0006057712836392856,
      "loss": 2.0653,
      "step": 980
    },
    {
      "epoch": 0.4497936726272352,
      "grad_norm": 0.21361298859119415,
      "learning_rate": 0.0006050452819736389,
      "loss": 2.0788,
      "step": 981
    },
    {
      "epoch": 0.45025217790004585,
      "grad_norm": 0.21120911836624146,
      "learning_rate": 0.000604319048538925,
      "loss": 2.0722,
      "step": 982
    },
    {
      "epoch": 0.4507106831728565,
      "grad_norm": 0.207930326461792,
      "learning_rate": 0.0006035925849374855,
      "loss": 2.0343,
      "step": 983
    },
    {
      "epoch": 0.4511691884456671,
      "grad_norm": 0.2078576534986496,
      "learning_rate": 0.0006028658927721697,
      "loss": 2.1013,
      "step": 984
    },
    {
      "epoch": 0.4516276937184778,
      "grad_norm": 0.19787511229515076,
      "learning_rate": 0.0006021389736463321,
      "loss": 2.0472,
      "step": 985
    },
    {
      "epoch": 0.4520861989912884,
      "grad_norm": 0.21289712190628052,
      "learning_rate": 0.0006014118291638271,
      "loss": 2.048,
      "step": 986
    },
    {
      "epoch": 0.45254470426409904,
      "grad_norm": 0.21783094108104706,
      "learning_rate": 0.0006006844609290065,
      "loss": 2.1001,
      "step": 987
    },
    {
      "epoch": 0.4530032095369097,
      "grad_norm": 0.20920640230178833,
      "learning_rate": 0.0005999568705467161,
      "loss": 2.081,
      "step": 988
    },
    {
      "epoch": 0.4534617148097203,
      "grad_norm": 0.2013660967350006,
      "learning_rate": 0.0005992290596222915,
      "loss": 2.0213,
      "step": 989
    },
    {
      "epoch": 0.45392022008253097,
      "grad_norm": 0.20473432540893555,
      "learning_rate": 0.0005985010297615551,
      "loss": 2.0013,
      "step": 990
    },
    {
      "epoch": 0.4543787253553416,
      "grad_norm": 0.22033101320266724,
      "learning_rate": 0.0005977727825708123,
      "loss": 2.0922,
      "step": 991
    },
    {
      "epoch": 0.45483723062815223,
      "grad_norm": 0.22145643830299377,
      "learning_rate": 0.0005970443196568478,
      "loss": 2.1013,
      "step": 992
    },
    {
      "epoch": 0.45529573590096284,
      "grad_norm": 0.22537536919116974,
      "learning_rate": 0.0005963156426269227,
      "loss": 2.1211,
      "step": 993
    },
    {
      "epoch": 0.4557542411737735,
      "grad_norm": 0.20744730532169342,
      "learning_rate": 0.0005955867530887702,
      "loss": 2.0266,
      "step": 994
    },
    {
      "epoch": 0.45621274644658416,
      "grad_norm": 0.20963235199451447,
      "learning_rate": 0.0005948576526505923,
      "loss": 2.0882,
      "step": 995
    },
    {
      "epoch": 0.45667125171939477,
      "grad_norm": 0.2130195051431656,
      "learning_rate": 0.0005941283429210568,
      "loss": 1.994,
      "step": 996
    },
    {
      "epoch": 0.4571297569922054,
      "grad_norm": 0.21371005475521088,
      "learning_rate": 0.0005933988255092926,
      "loss": 2.0817,
      "step": 997
    },
    {
      "epoch": 0.45758826226501603,
      "grad_norm": 0.205729678273201,
      "learning_rate": 0.0005926691020248874,
      "loss": 2.0506,
      "step": 998
    },
    {
      "epoch": 0.4580467675378267,
      "grad_norm": 0.21108320355415344,
      "learning_rate": 0.0005919391740778833,
      "loss": 2.0254,
      "step": 999
    },
    {
      "epoch": 0.4585052728106373,
      "grad_norm": 0.213540717959404,
      "learning_rate": 0.0005912090432787736,
      "loss": 2.0772,
      "step": 1000
    },
    {
      "epoch": 0.45896377808344796,
      "grad_norm": 0.2135593295097351,
      "learning_rate": 0.000590478711238499,
      "loss": 2.055,
      "step": 1001
    },
    {
      "epoch": 0.4594222833562586,
      "grad_norm": 0.21087217330932617,
      "learning_rate": 0.0005897481795684446,
      "loss": 2.0173,
      "step": 1002
    },
    {
      "epoch": 0.4598807886290692,
      "grad_norm": 0.210213765501976,
      "learning_rate": 0.0005890174498804355,
      "loss": 2.0922,
      "step": 1003
    },
    {
      "epoch": 0.4603392939018799,
      "grad_norm": 0.2012939155101776,
      "learning_rate": 0.0005882865237867339,
      "loss": 2.027,
      "step": 1004
    },
    {
      "epoch": 0.4607977991746905,
      "grad_norm": 0.19707202911376953,
      "learning_rate": 0.0005875554029000353,
      "loss": 2.0891,
      "step": 1005
    },
    {
      "epoch": 0.46125630444750115,
      "grad_norm": 0.19315926730632782,
      "learning_rate": 0.0005868240888334653,
      "loss": 2.0391,
      "step": 1006
    },
    {
      "epoch": 0.46171480972031176,
      "grad_norm": 0.20517626404762268,
      "learning_rate": 0.0005860925832005753,
      "loss": 2.1048,
      "step": 1007
    },
    {
      "epoch": 0.4621733149931224,
      "grad_norm": 0.21028241515159607,
      "learning_rate": 0.0005853608876153395,
      "loss": 2.1109,
      "step": 1008
    },
    {
      "epoch": 0.4626318202659331,
      "grad_norm": 0.20155684649944305,
      "learning_rate": 0.0005846290036921512,
      "loss": 2.018,
      "step": 1009
    },
    {
      "epoch": 0.4630903255387437,
      "grad_norm": 0.20317792892456055,
      "learning_rate": 0.0005838969330458195,
      "loss": 2.0199,
      "step": 1010
    },
    {
      "epoch": 0.46354883081155435,
      "grad_norm": 0.2090274840593338,
      "learning_rate": 0.0005831646772915651,
      "loss": 2.0259,
      "step": 1011
    },
    {
      "epoch": 0.46400733608436495,
      "grad_norm": 0.20905354619026184,
      "learning_rate": 0.0005824322380450173,
      "loss": 2.0673,
      "step": 1012
    },
    {
      "epoch": 0.4644658413571756,
      "grad_norm": 0.21790359914302826,
      "learning_rate": 0.0005816996169222102,
      "loss": 2.0465,
      "step": 1013
    },
    {
      "epoch": 0.4649243466299862,
      "grad_norm": 0.21703974902629852,
      "learning_rate": 0.0005809668155395793,
      "loss": 2.1131,
      "step": 1014
    },
    {
      "epoch": 0.4653828519027969,
      "grad_norm": 0.20858556032180786,
      "learning_rate": 0.0005802338355139578,
      "loss": 2.0412,
      "step": 1015
    },
    {
      "epoch": 0.46584135717560754,
      "grad_norm": 0.2066778540611267,
      "learning_rate": 0.0005795006784625728,
      "loss": 2.0922,
      "step": 1016
    },
    {
      "epoch": 0.46629986244841815,
      "grad_norm": 0.21842943131923676,
      "learning_rate": 0.0005787673460030423,
      "loss": 2.0946,
      "step": 1017
    },
    {
      "epoch": 0.4667583677212288,
      "grad_norm": 0.20547562837600708,
      "learning_rate": 0.000578033839753371,
      "loss": 2.0912,
      "step": 1018
    },
    {
      "epoch": 0.4672168729940394,
      "grad_norm": 0.21970736980438232,
      "learning_rate": 0.0005773001613319476,
      "loss": 1.9724,
      "step": 1019
    },
    {
      "epoch": 0.4676753782668501,
      "grad_norm": 0.21492122113704681,
      "learning_rate": 0.00057656631235754,
      "loss": 2.1501,
      "step": 1020
    },
    {
      "epoch": 0.46813388353966073,
      "grad_norm": 0.21355988085269928,
      "learning_rate": 0.0005758322944492929,
      "loss": 2.0842,
      "step": 1021
    },
    {
      "epoch": 0.46859238881247134,
      "grad_norm": 0.20416244864463806,
      "learning_rate": 0.0005750981092267237,
      "loss": 2.0634,
      "step": 1022
    },
    {
      "epoch": 0.469050894085282,
      "grad_norm": 0.20948675274848938,
      "learning_rate": 0.0005743637583097183,
      "loss": 2.0207,
      "step": 1023
    },
    {
      "epoch": 0.4695093993580926,
      "grad_norm": 0.2139819860458374,
      "learning_rate": 0.0005736292433185291,
      "loss": 2.0236,
      "step": 1024
    },
    {
      "epoch": 0.46996790463090327,
      "grad_norm": 0.2161521315574646,
      "learning_rate": 0.0005728945658737699,
      "loss": 2.1477,
      "step": 1025
    },
    {
      "epoch": 0.47042640990371387,
      "grad_norm": 0.20149926841259003,
      "learning_rate": 0.0005721597275964133,
      "loss": 2.0449,
      "step": 1026
    },
    {
      "epoch": 0.47088491517652453,
      "grad_norm": 0.21835586428642273,
      "learning_rate": 0.0005714247301077865,
      "loss": 2.0652,
      "step": 1027
    },
    {
      "epoch": 0.4713434204493352,
      "grad_norm": 0.19570407271385193,
      "learning_rate": 0.0005706895750295682,
      "loss": 2.1022,
      "step": 1028
    },
    {
      "epoch": 0.4718019257221458,
      "grad_norm": 0.2115260362625122,
      "learning_rate": 0.0005699542639837844,
      "loss": 2.0453,
      "step": 1029
    },
    {
      "epoch": 0.47226043099495646,
      "grad_norm": 0.20308710634708405,
      "learning_rate": 0.0005692187985928055,
      "loss": 2.1647,
      "step": 1030
    },
    {
      "epoch": 0.47271893626776706,
      "grad_norm": 0.20548376441001892,
      "learning_rate": 0.0005684831804793427,
      "loss": 2.244,
      "step": 1031
    },
    {
      "epoch": 0.4731774415405777,
      "grad_norm": 0.21307188272476196,
      "learning_rate": 0.0005677474112664438,
      "loss": 2.0033,
      "step": 1032
    },
    {
      "epoch": 0.47363594681338833,
      "grad_norm": 0.2303806096315384,
      "learning_rate": 0.0005670114925774899,
      "loss": 2.0782,
      "step": 1033
    },
    {
      "epoch": 0.474094452086199,
      "grad_norm": 0.20692144334316254,
      "learning_rate": 0.0005662754260361924,
      "loss": 2.0398,
      "step": 1034
    },
    {
      "epoch": 0.47455295735900965,
      "grad_norm": 0.21159429848194122,
      "learning_rate": 0.0005655392132665884,
      "loss": 2.0689,
      "step": 1035
    },
    {
      "epoch": 0.47501146263182026,
      "grad_norm": 0.21439136564731598,
      "learning_rate": 0.000564802855893038,
      "loss": 2.0433,
      "step": 1036
    },
    {
      "epoch": 0.4754699679046309,
      "grad_norm": 0.21965956687927246,
      "learning_rate": 0.0005640663555402198,
      "loss": 2.1396,
      "step": 1037
    },
    {
      "epoch": 0.4759284731774415,
      "grad_norm": 0.2099628746509552,
      "learning_rate": 0.0005633297138331285,
      "loss": 2.1344,
      "step": 1038
    },
    {
      "epoch": 0.4763869784502522,
      "grad_norm": 0.21291464567184448,
      "learning_rate": 0.0005625929323970705,
      "loss": 2.1838,
      "step": 1039
    },
    {
      "epoch": 0.4768454837230628,
      "grad_norm": 0.20207712054252625,
      "learning_rate": 0.0005618560128576603,
      "loss": 2.0634,
      "step": 1040
    },
    {
      "epoch": 0.47730398899587345,
      "grad_norm": 0.2175692915916443,
      "learning_rate": 0.0005611189568408173,
      "loss": 2.0838,
      "step": 1041
    },
    {
      "epoch": 0.4777624942686841,
      "grad_norm": 0.21294957399368286,
      "learning_rate": 0.0005603817659727619,
      "loss": 2.002,
      "step": 1042
    },
    {
      "epoch": 0.4782209995414947,
      "grad_norm": 0.21341069042682648,
      "learning_rate": 0.0005596444418800121,
      "loss": 2.0175,
      "step": 1043
    },
    {
      "epoch": 0.4786795048143054,
      "grad_norm": 0.21167002618312836,
      "learning_rate": 0.0005589069861893798,
      "loss": 2.0675,
      "step": 1044
    },
    {
      "epoch": 0.479138010087116,
      "grad_norm": 0.20804885029792786,
      "learning_rate": 0.0005581694005279673,
      "loss": 2.0599,
      "step": 1045
    },
    {
      "epoch": 0.47959651535992665,
      "grad_norm": 0.20882685482501984,
      "learning_rate": 0.0005574316865231637,
      "loss": 2.0146,
      "step": 1046
    },
    {
      "epoch": 0.48005502063273725,
      "grad_norm": 0.2154678851366043,
      "learning_rate": 0.0005566938458026411,
      "loss": 2.0229,
      "step": 1047
    },
    {
      "epoch": 0.4805135259055479,
      "grad_norm": 0.2074185609817505,
      "learning_rate": 0.0005559558799943514,
      "loss": 1.9927,
      "step": 1048
    },
    {
      "epoch": 0.48097203117835857,
      "grad_norm": 0.2152828723192215,
      "learning_rate": 0.0005552177907265223,
      "loss": 1.9205,
      "step": 1049
    },
    {
      "epoch": 0.4814305364511692,
      "grad_norm": 0.21146570146083832,
      "learning_rate": 0.000554479579627654,
      "loss": 2.0258,
      "step": 1050
    },
    {
      "epoch": 0.48188904172397984,
      "grad_norm": 0.23547549545764923,
      "learning_rate": 0.0005537412483265157,
      "loss": 2.1117,
      "step": 1051
    },
    {
      "epoch": 0.48234754699679044,
      "grad_norm": 0.22304974496364594,
      "learning_rate": 0.0005530027984521413,
      "loss": 2.0948,
      "step": 1052
    },
    {
      "epoch": 0.4828060522696011,
      "grad_norm": 0.2141544669866562,
      "learning_rate": 0.0005522642316338268,
      "loss": 2.0096,
      "step": 1053
    },
    {
      "epoch": 0.4832645575424117,
      "grad_norm": 0.21880905330181122,
      "learning_rate": 0.0005515255495011259,
      "loss": 2.0301,
      "step": 1054
    },
    {
      "epoch": 0.48372306281522237,
      "grad_norm": 0.21698138117790222,
      "learning_rate": 0.0005507867536838472,
      "loss": 2.0158,
      "step": 1055
    },
    {
      "epoch": 0.48418156808803303,
      "grad_norm": 0.21004213392734528,
      "learning_rate": 0.0005500478458120492,
      "loss": 2.0125,
      "step": 1056
    },
    {
      "epoch": 0.48464007336084364,
      "grad_norm": 0.20244596898555756,
      "learning_rate": 0.0005493088275160387,
      "loss": 2.0684,
      "step": 1057
    },
    {
      "epoch": 0.4850985786336543,
      "grad_norm": 0.2079736739397049,
      "learning_rate": 0.0005485697004263657,
      "loss": 2.0645,
      "step": 1058
    },
    {
      "epoch": 0.4855570839064649,
      "grad_norm": 0.20712174475193024,
      "learning_rate": 0.0005478304661738199,
      "loss": 2.1309,
      "step": 1059
    },
    {
      "epoch": 0.48601558917927556,
      "grad_norm": 0.21634802222251892,
      "learning_rate": 0.0005470911263894279,
      "loss": 2.0139,
      "step": 1060
    },
    {
      "epoch": 0.4864740944520862,
      "grad_norm": 0.2115454077720642,
      "learning_rate": 0.0005463516827044491,
      "loss": 2.057,
      "step": 1061
    },
    {
      "epoch": 0.48693259972489683,
      "grad_norm": 0.2079354226589203,
      "learning_rate": 0.000545612136750372,
      "loss": 2.0156,
      "step": 1062
    },
    {
      "epoch": 0.4873911049977075,
      "grad_norm": 0.19922232627868652,
      "learning_rate": 0.0005448724901589107,
      "loss": 2.1961,
      "step": 1063
    },
    {
      "epoch": 0.4878496102705181,
      "grad_norm": 0.20595909655094147,
      "learning_rate": 0.0005441327445620014,
      "loss": 2.0492,
      "step": 1064
    },
    {
      "epoch": 0.48830811554332876,
      "grad_norm": 0.221109077334404,
      "learning_rate": 0.0005433929015917988,
      "loss": 2.0114,
      "step": 1065
    },
    {
      "epoch": 0.48876662081613936,
      "grad_norm": 0.20459549129009247,
      "learning_rate": 0.0005426529628806724,
      "loss": 2.0054,
      "step": 1066
    },
    {
      "epoch": 0.48922512608895,
      "grad_norm": 0.21229512989521027,
      "learning_rate": 0.0005419129300612029,
      "loss": 2.0614,
      "step": 1067
    },
    {
      "epoch": 0.4896836313617607,
      "grad_norm": 0.20616242289543152,
      "learning_rate": 0.000541172804766179,
      "loss": 2.1332,
      "step": 1068
    },
    {
      "epoch": 0.4901421366345713,
      "grad_norm": 0.20714251697063446,
      "learning_rate": 0.0005404325886285927,
      "loss": 1.9954,
      "step": 1069
    },
    {
      "epoch": 0.49060064190738195,
      "grad_norm": 0.23401999473571777,
      "learning_rate": 0.000539692283281637,
      "loss": 2.0869,
      "step": 1070
    },
    {
      "epoch": 0.49105914718019256,
      "grad_norm": 0.21639320254325867,
      "learning_rate": 0.0005389518903587017,
      "loss": 2.0908,
      "step": 1071
    },
    {
      "epoch": 0.4915176524530032,
      "grad_norm": 0.21524940431118011,
      "learning_rate": 0.0005382114114933695,
      "loss": 2.0371,
      "step": 1072
    },
    {
      "epoch": 0.4919761577258138,
      "grad_norm": 0.2101580947637558,
      "learning_rate": 0.0005374708483194132,
      "loss": 2.0635,
      "step": 1073
    },
    {
      "epoch": 0.4924346629986245,
      "grad_norm": 0.22279341518878937,
      "learning_rate": 0.000536730202470791,
      "loss": 2.0421,
      "step": 1074
    },
    {
      "epoch": 0.49289316827143514,
      "grad_norm": 0.22329683601856232,
      "learning_rate": 0.0005359894755816443,
      "loss": 2.0411,
      "step": 1075
    },
    {
      "epoch": 0.49335167354424575,
      "grad_norm": 0.20261920988559723,
      "learning_rate": 0.0005352486692862926,
      "loss": 2.079,
      "step": 1076
    },
    {
      "epoch": 0.4938101788170564,
      "grad_norm": 0.2070843130350113,
      "learning_rate": 0.0005345077852192307,
      "loss": 2.1129,
      "step": 1077
    },
    {
      "epoch": 0.494268684089867,
      "grad_norm": 0.21265257894992828,
      "learning_rate": 0.0005337668250151254,
      "loss": 2.0441,
      "step": 1078
    },
    {
      "epoch": 0.4947271893626777,
      "grad_norm": 0.2076832354068756,
      "learning_rate": 0.0005330257903088111,
      "loss": 2.0184,
      "step": 1079
    },
    {
      "epoch": 0.4951856946354883,
      "grad_norm": 0.22389812767505646,
      "learning_rate": 0.000532284682735287,
      "loss": 1.9981,
      "step": 1080
    },
    {
      "epoch": 0.49564419990829894,
      "grad_norm": 0.20203141868114471,
      "learning_rate": 0.0005315435039297124,
      "loss": 2.0591,
      "step": 1081
    },
    {
      "epoch": 0.4961027051811096,
      "grad_norm": 0.1999691277742386,
      "learning_rate": 0.0005308022555274046,
      "loss": 2.0115,
      "step": 1082
    },
    {
      "epoch": 0.4965612104539202,
      "grad_norm": 0.2016538381576538,
      "learning_rate": 0.0005300609391638336,
      "loss": 2.0416,
      "step": 1083
    },
    {
      "epoch": 0.49701971572673087,
      "grad_norm": 0.19702544808387756,
      "learning_rate": 0.0005293195564746201,
      "loss": 2.0203,
      "step": 1084
    },
    {
      "epoch": 0.4974782209995415,
      "grad_norm": 0.20484475791454315,
      "learning_rate": 0.0005285781090955304,
      "loss": 2.0222,
      "step": 1085
    },
    {
      "epoch": 0.49793672627235214,
      "grad_norm": 0.21232926845550537,
      "learning_rate": 0.0005278365986624743,
      "loss": 2.0169,
      "step": 1086
    },
    {
      "epoch": 0.49839523154516274,
      "grad_norm": 0.2048821747303009,
      "learning_rate": 0.0005270950268115001,
      "loss": 2.0158,
      "step": 1087
    },
    {
      "epoch": 0.4988537368179734,
      "grad_norm": 0.20197269320487976,
      "learning_rate": 0.0005263533951787919,
      "loss": 2.0705,
      "step": 1088
    },
    {
      "epoch": 0.49931224209078406,
      "grad_norm": 0.2156265527009964,
      "learning_rate": 0.000525611705400666,
      "loss": 2.0977,
      "step": 1089
    },
    {
      "epoch": 0.49977074736359467,
      "grad_norm": 0.19378478825092316,
      "learning_rate": 0.0005248699591135664,
      "loss": 2.0576,
      "step": 1090
    },
    {
      "epoch": 0.5002292526364053,
      "grad_norm": 0.2012927532196045,
      "learning_rate": 0.0005241281579540618,
      "loss": 2.0123,
      "step": 1091
    },
    {
      "epoch": 0.5006877579092159,
      "grad_norm": 0.20989751815795898,
      "learning_rate": 0.0005233863035588427,
      "loss": 2.1397,
      "step": 1092
    },
    {
      "epoch": 0.5011462631820266,
      "grad_norm": 0.20942288637161255,
      "learning_rate": 0.0005226443975647161,
      "loss": 2.1101,
      "step": 1093
    },
    {
      "epoch": 0.5016047684548373,
      "grad_norm": 0.2150336503982544,
      "learning_rate": 0.0005219024416086036,
      "loss": 2.0245,
      "step": 1094
    },
    {
      "epoch": 0.5020632737276479,
      "grad_norm": 0.20209649205207825,
      "learning_rate": 0.0005211604373275366,
      "loss": 2.0852,
      "step": 1095
    },
    {
      "epoch": 0.5025217790004585,
      "grad_norm": 0.21479089558124542,
      "learning_rate": 0.0005204183863586533,
      "loss": 1.9797,
      "step": 1096
    },
    {
      "epoch": 0.5029802842732691,
      "grad_norm": 0.21780915558338165,
      "learning_rate": 0.0005196762903391951,
      "loss": 2.0809,
      "step": 1097
    },
    {
      "epoch": 0.5034387895460798,
      "grad_norm": 0.20210231840610504,
      "learning_rate": 0.0005189341509065023,
      "loss": 2.0672,
      "step": 1098
    },
    {
      "epoch": 0.5038972948188905,
      "grad_norm": 0.2125033289194107,
      "learning_rate": 0.0005181919696980112,
      "loss": 2.1085,
      "step": 1099
    },
    {
      "epoch": 0.504355800091701,
      "grad_norm": 0.20507454872131348,
      "learning_rate": 0.0005174497483512506,
      "loss": 2.0713,
      "step": 1100
    },
    {
      "epoch": 0.5048143053645117,
      "grad_norm": 0.21028780937194824,
      "learning_rate": 0.0005167074885038374,
      "loss": 1.9714,
      "step": 1101
    },
    {
      "epoch": 0.5052728106373223,
      "grad_norm": 0.31266850233078003,
      "learning_rate": 0.0005159651917934735,
      "loss": 2.0256,
      "step": 1102
    },
    {
      "epoch": 0.505731315910133,
      "grad_norm": 0.21719002723693848,
      "learning_rate": 0.0005152228598579428,
      "loss": 2.1167,
      "step": 1103
    },
    {
      "epoch": 0.5061898211829436,
      "grad_norm": 0.20560823380947113,
      "learning_rate": 0.000514480494335106,
      "loss": 2.0254,
      "step": 1104
    },
    {
      "epoch": 0.5066483264557542,
      "grad_norm": 0.19843913614749908,
      "learning_rate": 0.0005137380968628983,
      "loss": 2.072,
      "step": 1105
    },
    {
      "epoch": 0.5071068317285649,
      "grad_norm": 0.20444811880588531,
      "learning_rate": 0.0005129956690793255,
      "loss": 2.0167,
      "step": 1106
    },
    {
      "epoch": 0.5075653370013755,
      "grad_norm": 0.2037172019481659,
      "learning_rate": 0.0005122532126224601,
      "loss": 2.1246,
      "step": 1107
    },
    {
      "epoch": 0.5080238422741862,
      "grad_norm": 0.22151310741901398,
      "learning_rate": 0.0005115107291304378,
      "loss": 2.0818,
      "step": 1108
    },
    {
      "epoch": 0.5084823475469968,
      "grad_norm": 0.2020580768585205,
      "learning_rate": 0.0005107682202414544,
      "loss": 2.031,
      "step": 1109
    },
    {
      "epoch": 0.5089408528198074,
      "grad_norm": 0.2008032500743866,
      "learning_rate": 0.0005100256875937613,
      "loss": 2.0054,
      "step": 1110
    },
    {
      "epoch": 0.509399358092618,
      "grad_norm": 0.21867255866527557,
      "learning_rate": 0.0005092831328256625,
      "loss": 2.097,
      "step": 1111
    },
    {
      "epoch": 0.5098578633654287,
      "grad_norm": 0.20906470715999603,
      "learning_rate": 0.0005085405575755105,
      "loss": 2.0871,
      "step": 1112
    },
    {
      "epoch": 0.5103163686382394,
      "grad_norm": 0.20772957801818848,
      "learning_rate": 0.0005077979634817034,
      "loss": 1.9869,
      "step": 1113
    },
    {
      "epoch": 0.51077487391105,
      "grad_norm": 0.2033894956111908,
      "learning_rate": 0.0005070553521826808,
      "loss": 1.9805,
      "step": 1114
    },
    {
      "epoch": 0.5112333791838606,
      "grad_norm": 0.20701146125793457,
      "learning_rate": 0.00050631272531692,
      "loss": 2.1301,
      "step": 1115
    },
    {
      "epoch": 0.5116918844566712,
      "grad_norm": 0.20044931769371033,
      "learning_rate": 0.0005055700845229327,
      "loss": 2.0537,
      "step": 1116
    },
    {
      "epoch": 0.5121503897294819,
      "grad_norm": 0.2021133154630661,
      "learning_rate": 0.000504827431439262,
      "loss": 2.0845,
      "step": 1117
    },
    {
      "epoch": 0.5126088950022926,
      "grad_norm": 0.19496183097362518,
      "learning_rate": 0.000504084767704477,
      "loss": 2.0289,
      "step": 1118
    },
    {
      "epoch": 0.5130674002751031,
      "grad_norm": 0.20649433135986328,
      "learning_rate": 0.0005033420949571712,
      "loss": 1.9924,
      "step": 1119
    },
    {
      "epoch": 0.5135259055479138,
      "grad_norm": 0.20320338010787964,
      "learning_rate": 0.0005025994148359574,
      "loss": 2.0409,
      "step": 1120
    },
    {
      "epoch": 0.5139844108207244,
      "grad_norm": 0.21451222896575928,
      "learning_rate": 0.0005018567289794651,
      "loss": 1.95,
      "step": 1121
    },
    {
      "epoch": 0.5144429160935351,
      "grad_norm": 0.21378003060817719,
      "learning_rate": 0.0005011140390263362,
      "loss": 2.1079,
      "step": 1122
    },
    {
      "epoch": 0.5149014213663458,
      "grad_norm": 0.2186053991317749,
      "learning_rate": 0.0005003713466152218,
      "loss": 2.0642,
      "step": 1123
    },
    {
      "epoch": 0.5153599266391563,
      "grad_norm": 0.20801058411598206,
      "learning_rate": 0.0004996286533847783,
      "loss": 2.0541,
      "step": 1124
    },
    {
      "epoch": 0.515818431911967,
      "grad_norm": 0.21296656131744385,
      "learning_rate": 0.000498885960973664,
      "loss": 2.0782,
      "step": 1125
    },
    {
      "epoch": 0.5162769371847776,
      "grad_norm": 0.22307997941970825,
      "learning_rate": 0.000498143271020535,
      "loss": 2.0116,
      "step": 1126
    },
    {
      "epoch": 0.5167354424575883,
      "grad_norm": 0.22187963128089905,
      "learning_rate": 0.0004974005851640428,
      "loss": 2.1043,
      "step": 1127
    },
    {
      "epoch": 0.517193947730399,
      "grad_norm": 0.20855902135372162,
      "learning_rate": 0.000496657905042829,
      "loss": 2.0499,
      "step": 1128
    },
    {
      "epoch": 0.5176524530032095,
      "grad_norm": 0.20718683302402496,
      "learning_rate": 0.0004959152322955232,
      "loss": 2.0211,
      "step": 1129
    },
    {
      "epoch": 0.5181109582760202,
      "grad_norm": 0.21088528633117676,
      "learning_rate": 0.0004951725685607382,
      "loss": 2.1461,
      "step": 1130
    },
    {
      "epoch": 0.5185694635488308,
      "grad_norm": 0.21637214720249176,
      "learning_rate": 0.0004944299154770673,
      "loss": 2.0123,
      "step": 1131
    },
    {
      "epoch": 0.5190279688216415,
      "grad_norm": 0.21082259714603424,
      "learning_rate": 0.0004936872746830802,
      "loss": 2.0329,
      "step": 1132
    },
    {
      "epoch": 0.519486474094452,
      "grad_norm": 0.20381450653076172,
      "learning_rate": 0.0004929446478173195,
      "loss": 2.0496,
      "step": 1133
    },
    {
      "epoch": 0.5199449793672627,
      "grad_norm": 0.21353910863399506,
      "learning_rate": 0.0004922020365182968,
      "loss": 2.0081,
      "step": 1134
    },
    {
      "epoch": 0.5204034846400734,
      "grad_norm": 0.20848211646080017,
      "learning_rate": 0.0004914594424244897,
      "loss": 2.0187,
      "step": 1135
    },
    {
      "epoch": 0.520861989912884,
      "grad_norm": 0.21212221682071686,
      "learning_rate": 0.0004907168671743376,
      "loss": 2.0158,
      "step": 1136
    },
    {
      "epoch": 0.5213204951856947,
      "grad_norm": 0.20360974967479706,
      "learning_rate": 0.0004899743124062387,
      "loss": 1.9852,
      "step": 1137
    },
    {
      "epoch": 0.5217790004585052,
      "grad_norm": 0.2175011783838272,
      "learning_rate": 0.0004892317797585456,
      "loss": 2.0092,
      "step": 1138
    },
    {
      "epoch": 0.5222375057313159,
      "grad_norm": 0.21263253688812256,
      "learning_rate": 0.0004884892708695623,
      "loss": 1.9717,
      "step": 1139
    },
    {
      "epoch": 0.5226960110041265,
      "grad_norm": 0.22256921231746674,
      "learning_rate": 0.0004877467873775402,
      "loss": 2.0621,
      "step": 1140
    },
    {
      "epoch": 0.5231545162769372,
      "grad_norm": 0.20930485427379608,
      "learning_rate": 0.00048700433092067473,
      "loss": 2.098,
      "step": 1141
    },
    {
      "epoch": 0.5236130215497479,
      "grad_norm": 0.20646582543849945,
      "learning_rate": 0.0004862619031371019,
      "loss": 2.0413,
      "step": 1142
    },
    {
      "epoch": 0.5240715268225584,
      "grad_norm": 0.20976178348064423,
      "learning_rate": 0.0004855195056648942,
      "loss": 1.9947,
      "step": 1143
    },
    {
      "epoch": 0.5245300320953691,
      "grad_norm": 0.2024880200624466,
      "learning_rate": 0.00048477714014205734,
      "loss": 2.0411,
      "step": 1144
    },
    {
      "epoch": 0.5249885373681797,
      "grad_norm": 0.20159803330898285,
      "learning_rate": 0.00048403480820652644,
      "loss": 2.1066,
      "step": 1145
    },
    {
      "epoch": 0.5254470426409904,
      "grad_norm": 0.19967858493328094,
      "learning_rate": 0.0004832925114961629,
      "loss": 2.0681,
      "step": 1146
    },
    {
      "epoch": 0.5259055479138011,
      "grad_norm": 0.19834791123867035,
      "learning_rate": 0.0004825502516487497,
      "loss": 2.0719,
      "step": 1147
    },
    {
      "epoch": 0.5263640531866116,
      "grad_norm": 0.20275187492370605,
      "learning_rate": 0.00048180803030198896,
      "loss": 2.1123,
      "step": 1148
    },
    {
      "epoch": 0.5268225584594223,
      "grad_norm": 0.21217948198318481,
      "learning_rate": 0.0004810658490934979,
      "loss": 2.0816,
      "step": 1149
    },
    {
      "epoch": 0.5272810637322329,
      "grad_norm": 0.20592071115970612,
      "learning_rate": 0.000480323709660805,
      "loss": 2.1197,
      "step": 1150
    },
    {
      "epoch": 0.5277395690050436,
      "grad_norm": 0.2153848260641098,
      "learning_rate": 0.0004795816136413467,
      "loss": 1.9653,
      "step": 1151
    },
    {
      "epoch": 0.5281980742778541,
      "grad_norm": 0.20051167905330658,
      "learning_rate": 0.00047883956267246353,
      "loss": 2.0812,
      "step": 1152
    },
    {
      "epoch": 0.5286565795506648,
      "grad_norm": 0.20217469334602356,
      "learning_rate": 0.00047809755839139657,
      "loss": 2.0933,
      "step": 1153
    },
    {
      "epoch": 0.5291150848234755,
      "grad_norm": 0.2128201127052307,
      "learning_rate": 0.0004773556024352841,
      "loss": 2.0943,
      "step": 1154
    },
    {
      "epoch": 0.5295735900962861,
      "grad_norm": 0.19801776111125946,
      "learning_rate": 0.00047661369644115754,
      "loss": 1.94,
      "step": 1155
    },
    {
      "epoch": 0.5300320953690968,
      "grad_norm": 0.20083217322826385,
      "learning_rate": 0.0004758718420459383,
      "loss": 1.9593,
      "step": 1156
    },
    {
      "epoch": 0.5304906006419073,
      "grad_norm": 0.20288150012493134,
      "learning_rate": 0.0004751300408864339,
      "loss": 1.9769,
      "step": 1157
    },
    {
      "epoch": 0.530949105914718,
      "grad_norm": 0.21859005093574524,
      "learning_rate": 0.00047438829459933414,
      "loss": 2.0555,
      "step": 1158
    },
    {
      "epoch": 0.5314076111875287,
      "grad_norm": 0.21503470838069916,
      "learning_rate": 0.0004736466048212082,
      "loss": 2.0921,
      "step": 1159
    },
    {
      "epoch": 0.5318661164603393,
      "grad_norm": 0.21432562172412872,
      "learning_rate": 0.0004729049731885002,
      "loss": 2.0201,
      "step": 1160
    },
    {
      "epoch": 0.53232462173315,
      "grad_norm": 0.2015000581741333,
      "learning_rate": 0.000472163401337526,
      "loss": 2.0207,
      "step": 1161
    },
    {
      "epoch": 0.5327831270059605,
      "grad_norm": 0.19777435064315796,
      "learning_rate": 0.00047142189090446985,
      "loss": 1.993,
      "step": 1162
    },
    {
      "epoch": 0.5332416322787712,
      "grad_norm": 0.2004212588071823,
      "learning_rate": 0.0004706804435253802,
      "loss": 2.1249,
      "step": 1163
    },
    {
      "epoch": 0.5337001375515819,
      "grad_norm": 0.21052177250385284,
      "learning_rate": 0.0004699390608361665,
      "loss": 2.1436,
      "step": 1164
    },
    {
      "epoch": 0.5341586428243925,
      "grad_norm": 0.21137721836566925,
      "learning_rate": 0.0004691977444725955,
      "loss": 2.0534,
      "step": 1165
    },
    {
      "epoch": 0.5346171480972031,
      "grad_norm": 0.21830375492572784,
      "learning_rate": 0.0004684564960702877,
      "loss": 2.0634,
      "step": 1166
    },
    {
      "epoch": 0.5350756533700137,
      "grad_norm": 0.2094404101371765,
      "learning_rate": 0.0004677153172647131,
      "loss": 2.0663,
      "step": 1167
    },
    {
      "epoch": 0.5355341586428244,
      "grad_norm": 0.21606306731700897,
      "learning_rate": 0.00046697420969118894,
      "loss": 2.1029,
      "step": 1168
    },
    {
      "epoch": 0.535992663915635,
      "grad_norm": 0.21189160645008087,
      "learning_rate": 0.00046623317498487466,
      "loss": 2.0103,
      "step": 1169
    },
    {
      "epoch": 0.5364511691884457,
      "grad_norm": 0.2216084599494934,
      "learning_rate": 0.0004654922147807694,
      "loss": 2.1246,
      "step": 1170
    },
    {
      "epoch": 0.5369096744612563,
      "grad_norm": 0.19631634652614594,
      "learning_rate": 0.00046475133071370757,
      "loss": 1.981,
      "step": 1171
    },
    {
      "epoch": 0.5373681797340669,
      "grad_norm": 0.21219474077224731,
      "learning_rate": 0.00046401052441835574,
      "loss": 2.0573,
      "step": 1172
    },
    {
      "epoch": 0.5378266850068776,
      "grad_norm": 0.2150328904390335,
      "learning_rate": 0.000463269797529209,
      "loss": 1.9234,
      "step": 1173
    },
    {
      "epoch": 0.5382851902796882,
      "grad_norm": 0.2096128761768341,
      "learning_rate": 0.00046252915168058697,
      "loss": 2.0899,
      "step": 1174
    },
    {
      "epoch": 0.5387436955524989,
      "grad_norm": 0.21790243685245514,
      "learning_rate": 0.0004617885885066305,
      "loss": 1.9955,
      "step": 1175
    },
    {
      "epoch": 0.5392022008253095,
      "grad_norm": 0.22332757711410522,
      "learning_rate": 0.0004610481096412984,
      "loss": 2.0645,
      "step": 1176
    },
    {
      "epoch": 0.5396607060981201,
      "grad_norm": 0.21513494849205017,
      "learning_rate": 0.000460307716718363,
      "loss": 2.0518,
      "step": 1177
    },
    {
      "epoch": 0.5401192113709308,
      "grad_norm": 0.21508604288101196,
      "learning_rate": 0.0004595674113714074,
      "loss": 2.0117,
      "step": 1178
    },
    {
      "epoch": 0.5405777166437414,
      "grad_norm": 0.1992771327495575,
      "learning_rate": 0.0004588271952338212,
      "loss": 2.0699,
      "step": 1179
    },
    {
      "epoch": 0.5410362219165521,
      "grad_norm": 0.21164365112781525,
      "learning_rate": 0.00045808706993879714,
      "loss": 2.1452,
      "step": 1180
    },
    {
      "epoch": 0.5414947271893626,
      "grad_norm": 0.20152069628238678,
      "learning_rate": 0.00045734703711932767,
      "loss": 2.0622,
      "step": 1181
    },
    {
      "epoch": 0.5419532324621733,
      "grad_norm": 0.19984497129917145,
      "learning_rate": 0.0004566070984082013,
      "loss": 1.9732,
      "step": 1182
    },
    {
      "epoch": 0.542411737734984,
      "grad_norm": 0.19975166022777557,
      "learning_rate": 0.00045586725543799865,
      "loss": 1.9804,
      "step": 1183
    },
    {
      "epoch": 0.5428702430077946,
      "grad_norm": 0.20334453880786896,
      "learning_rate": 0.00045512750984108937,
      "loss": 2.0916,
      "step": 1184
    },
    {
      "epoch": 0.5433287482806052,
      "grad_norm": 0.20853489637374878,
      "learning_rate": 0.000454387863249628,
      "loss": 2.1325,
      "step": 1185
    },
    {
      "epoch": 0.5437872535534158,
      "grad_norm": 0.20991018414497375,
      "learning_rate": 0.00045364831729555096,
      "loss": 2.0375,
      "step": 1186
    },
    {
      "epoch": 0.5442457588262265,
      "grad_norm": 0.21497392654418945,
      "learning_rate": 0.0004529088736105721,
      "loss": 2.0849,
      "step": 1187
    },
    {
      "epoch": 0.5447042640990372,
      "grad_norm": 0.2056678831577301,
      "learning_rate": 0.0004521695338261802,
      "loss": 2.092,
      "step": 1188
    },
    {
      "epoch": 0.5451627693718478,
      "grad_norm": 0.21992243826389313,
      "learning_rate": 0.0004514302995736344,
      "loss": 2.0166,
      "step": 1189
    },
    {
      "epoch": 0.5456212746446584,
      "grad_norm": 0.20607493817806244,
      "learning_rate": 0.0004506911724839613,
      "loss": 2.0533,
      "step": 1190
    },
    {
      "epoch": 0.546079779917469,
      "grad_norm": 0.20113621652126312,
      "learning_rate": 0.0004499521541879508,
      "loss": 1.9707,
      "step": 1191
    },
    {
      "epoch": 0.5465382851902797,
      "grad_norm": 0.2118975967168808,
      "learning_rate": 0.00044921324631615303,
      "loss": 2.0327,
      "step": 1192
    },
    {
      "epoch": 0.5469967904630904,
      "grad_norm": 0.19567330181598663,
      "learning_rate": 0.0004484744504988742,
      "loss": 2.0223,
      "step": 1193
    },
    {
      "epoch": 0.547455295735901,
      "grad_norm": 0.2181934118270874,
      "learning_rate": 0.00044773576836617336,
      "loss": 2.0213,
      "step": 1194
    },
    {
      "epoch": 0.5479138010087116,
      "grad_norm": 0.2068590521812439,
      "learning_rate": 0.0004469972015478588,
      "loss": 2.0439,
      "step": 1195
    },
    {
      "epoch": 0.5483723062815222,
      "grad_norm": 0.20782645046710968,
      "learning_rate": 0.0004462587516734844,
      "loss": 2.0253,
      "step": 1196
    },
    {
      "epoch": 0.5488308115543329,
      "grad_norm": 0.2099522203207016,
      "learning_rate": 0.00044552042037234596,
      "loss": 2.0553,
      "step": 1197
    },
    {
      "epoch": 0.5492893168271435,
      "grad_norm": 0.20520374178886414,
      "learning_rate": 0.00044478220927347774,
      "loss": 2.0422,
      "step": 1198
    },
    {
      "epoch": 0.5497478220999541,
      "grad_norm": 0.2042413055896759,
      "learning_rate": 0.00044404412000564875,
      "loss": 2.004,
      "step": 1199
    },
    {
      "epoch": 0.5502063273727648,
      "grad_norm": 0.20612218976020813,
      "learning_rate": 0.000443306154197359,
      "loss": 2.0347,
      "step": 1200
    },
    {
      "epoch": 0.5506648326455754,
      "grad_norm": 0.21076732873916626,
      "learning_rate": 0.00044256831347683646,
      "loss": 1.9993,
      "step": 1201
    },
    {
      "epoch": 0.5511233379183861,
      "grad_norm": 0.23036275804042816,
      "learning_rate": 0.0004418305994720328,
      "loss": 2.0334,
      "step": 1202
    },
    {
      "epoch": 0.5515818431911967,
      "grad_norm": 0.21030102670192719,
      "learning_rate": 0.0004410930138106203,
      "loss": 2.0408,
      "step": 1203
    },
    {
      "epoch": 0.5520403484640073,
      "grad_norm": 0.19949698448181152,
      "learning_rate": 0.000440355558119988,
      "loss": 2.0973,
      "step": 1204
    },
    {
      "epoch": 0.552498853736818,
      "grad_norm": 0.20909160375595093,
      "learning_rate": 0.00043961823402723814,
      "loss": 2.0223,
      "step": 1205
    },
    {
      "epoch": 0.5529573590096286,
      "grad_norm": 0.20901232957839966,
      "learning_rate": 0.0004388810431591829,
      "loss": 2.1169,
      "step": 1206
    },
    {
      "epoch": 0.5534158642824393,
      "grad_norm": 0.21776142716407776,
      "learning_rate": 0.0004381439871423398,
      "loss": 2.0326,
      "step": 1207
    },
    {
      "epoch": 0.5538743695552499,
      "grad_norm": 0.21298226714134216,
      "learning_rate": 0.00043740706760292966,
      "loss": 2.0179,
      "step": 1208
    },
    {
      "epoch": 0.5543328748280605,
      "grad_norm": 0.20370543003082275,
      "learning_rate": 0.0004366702861668716,
      "loss": 2.0534,
      "step": 1209
    },
    {
      "epoch": 0.5547913801008711,
      "grad_norm": 0.2028646022081375,
      "learning_rate": 0.00043593364445978036,
      "loss": 2.0678,
      "step": 1210
    },
    {
      "epoch": 0.5552498853736818,
      "grad_norm": 0.21842916309833527,
      "learning_rate": 0.0004351971441069622,
      "loss": 1.9479,
      "step": 1211
    },
    {
      "epoch": 0.5557083906464925,
      "grad_norm": 0.2075813114643097,
      "learning_rate": 0.0004344607867334116,
      "loss": 1.9855,
      "step": 1212
    },
    {
      "epoch": 0.556166895919303,
      "grad_norm": 0.20478644967079163,
      "learning_rate": 0.00043372457396380766,
      "loss": 1.9583,
      "step": 1213
    },
    {
      "epoch": 0.5566254011921137,
      "grad_norm": 0.2411949187517166,
      "learning_rate": 0.00043298850742251013,
      "loss": 2.1976,
      "step": 1214
    },
    {
      "epoch": 0.5570839064649243,
      "grad_norm": 0.21841974556446075,
      "learning_rate": 0.0004322525887335563,
      "loss": 2.0049,
      "step": 1215
    },
    {
      "epoch": 0.557542411737735,
      "grad_norm": 0.22550590336322784,
      "learning_rate": 0.00043151681952065734,
      "loss": 2.1122,
      "step": 1216
    },
    {
      "epoch": 0.5580009170105457,
      "grad_norm": 0.22114279866218567,
      "learning_rate": 0.00043078120140719456,
      "loss": 2.1057,
      "step": 1217
    },
    {
      "epoch": 0.5584594222833562,
      "grad_norm": 0.21660995483398438,
      "learning_rate": 0.0004300457360162158,
      "loss": 2.0429,
      "step": 1218
    },
    {
      "epoch": 0.5589179275561669,
      "grad_norm": 0.2117169052362442,
      "learning_rate": 0.0004293104249704319,
      "loss": 1.9979,
      "step": 1219
    },
    {
      "epoch": 0.5593764328289775,
      "grad_norm": 0.20704205334186554,
      "learning_rate": 0.00042857526989221355,
      "loss": 2.0815,
      "step": 1220
    },
    {
      "epoch": 0.5598349381017882,
      "grad_norm": 0.2117859274148941,
      "learning_rate": 0.00042784027240358674,
      "loss": 2.0407,
      "step": 1221
    },
    {
      "epoch": 0.5602934433745989,
      "grad_norm": 0.21628054976463318,
      "learning_rate": 0.0004271054341262301,
      "loss": 2.0105,
      "step": 1222
    },
    {
      "epoch": 0.5607519486474094,
      "grad_norm": 0.1993594467639923,
      "learning_rate": 0.000426370756681471,
      "loss": 2.0131,
      "step": 1223
    },
    {
      "epoch": 0.5612104539202201,
      "grad_norm": 0.2081214338541031,
      "learning_rate": 0.0004256362416902817,
      "loss": 2.0603,
      "step": 1224
    },
    {
      "epoch": 0.5616689591930307,
      "grad_norm": 0.20900852978229523,
      "learning_rate": 0.00042490189077327637,
      "loss": 2.0187,
      "step": 1225
    },
    {
      "epoch": 0.5621274644658414,
      "grad_norm": 0.21789489686489105,
      "learning_rate": 0.00042416770555070703,
      "loss": 2.0395,
      "step": 1226
    },
    {
      "epoch": 0.562585969738652,
      "grad_norm": 0.21205563843250275,
      "learning_rate": 0.00042343368764246,
      "loss": 1.9298,
      "step": 1227
    },
    {
      "epoch": 0.5630444750114626,
      "grad_norm": 0.21730759739875793,
      "learning_rate": 0.0004226998386680524,
      "loss": 2.1103,
      "step": 1228
    },
    {
      "epoch": 0.5635029802842733,
      "grad_norm": 0.20455501973628998,
      "learning_rate": 0.000421966160246629,
      "loss": 2.1163,
      "step": 1229
    },
    {
      "epoch": 0.5639614855570839,
      "grad_norm": 0.21531540155410767,
      "learning_rate": 0.00042123265399695783,
      "loss": 2.0275,
      "step": 1230
    },
    {
      "epoch": 0.5644199908298946,
      "grad_norm": 0.21708457171916962,
      "learning_rate": 0.0004204993215374273,
      "loss": 2.1931,
      "step": 1231
    },
    {
      "epoch": 0.5648784961027051,
      "grad_norm": 0.20668213069438934,
      "learning_rate": 0.00041976616448604226,
      "loss": 2.0842,
      "step": 1232
    },
    {
      "epoch": 0.5653370013755158,
      "grad_norm": 0.202859029173851,
      "learning_rate": 0.00041903318446042076,
      "loss": 2.0586,
      "step": 1233
    },
    {
      "epoch": 0.5657955066483265,
      "grad_norm": 0.19746744632720947,
      "learning_rate": 0.00041830038307778984,
      "loss": 2.11,
      "step": 1234
    },
    {
      "epoch": 0.5662540119211371,
      "grad_norm": 0.22127456963062286,
      "learning_rate": 0.0004175677619549828,
      "loss": 2.1123,
      "step": 1235
    },
    {
      "epoch": 0.5667125171939478,
      "grad_norm": 0.20658601820468903,
      "learning_rate": 0.000416835322708435,
      "loss": 1.9623,
      "step": 1236
    },
    {
      "epoch": 0.5671710224667583,
      "grad_norm": 0.2090967297554016,
      "learning_rate": 0.00041610306695418056,
      "loss": 2.0039,
      "step": 1237
    },
    {
      "epoch": 0.567629527739569,
      "grad_norm": 0.22067339718341827,
      "learning_rate": 0.0004153709963078488,
      "loss": 2.1005,
      "step": 1238
    },
    {
      "epoch": 0.5680880330123796,
      "grad_norm": 0.2140693962574005,
      "learning_rate": 0.0004146391123846606,
      "loss": 2.0521,
      "step": 1239
    },
    {
      "epoch": 0.5685465382851903,
      "grad_norm": 0.20408332347869873,
      "learning_rate": 0.0004139074167994249,
      "loss": 2.0427,
      "step": 1240
    },
    {
      "epoch": 0.569005043558001,
      "grad_norm": 0.2064693719148636,
      "learning_rate": 0.00041317591116653486,
      "loss": 2.0717,
      "step": 1241
    },
    {
      "epoch": 0.5694635488308115,
      "grad_norm": 0.19684873521327972,
      "learning_rate": 0.0004124445970999648,
      "loss": 2.0505,
      "step": 1242
    },
    {
      "epoch": 0.5699220541036222,
      "grad_norm": 0.21251673996448517,
      "learning_rate": 0.00041171347621326627,
      "loss": 2.0148,
      "step": 1243
    },
    {
      "epoch": 0.5703805593764328,
      "grad_norm": 0.20654599368572235,
      "learning_rate": 0.00041098255011956465,
      "loss": 2.1133,
      "step": 1244
    },
    {
      "epoch": 0.5708390646492435,
      "grad_norm": 0.20567791163921356,
      "learning_rate": 0.00041025182043155547,
      "loss": 2.0856,
      "step": 1245
    },
    {
      "epoch": 0.571297569922054,
      "grad_norm": 0.2100348323583603,
      "learning_rate": 0.000409521288761501,
      "loss": 2.0809,
      "step": 1246
    },
    {
      "epoch": 0.5717560751948647,
      "grad_norm": 0.20774655044078827,
      "learning_rate": 0.00040879095672122646,
      "loss": 2.0138,
      "step": 1247
    },
    {
      "epoch": 0.5722145804676754,
      "grad_norm": 0.2196817845106125,
      "learning_rate": 0.0004080608259221167,
      "loss": 2.0431,
      "step": 1248
    },
    {
      "epoch": 0.572673085740486,
      "grad_norm": 0.20007023215293884,
      "learning_rate": 0.0004073308979751126,
      "loss": 1.9926,
      "step": 1249
    },
    {
      "epoch": 0.5731315910132967,
      "grad_norm": 0.20421606302261353,
      "learning_rate": 0.0004066011744907074,
      "loss": 1.9502,
      "step": 1250
    },
    {
      "epoch": 0.5735900962861072,
      "grad_norm": 0.21492214500904083,
      "learning_rate": 0.00040587165707894326,
      "loss": 2.0775,
      "step": 1251
    },
    {
      "epoch": 0.5740486015589179,
      "grad_norm": 0.20391112565994263,
      "learning_rate": 0.0004051423473494076,
      "loss": 2.0039,
      "step": 1252
    },
    {
      "epoch": 0.5745071068317286,
      "grad_norm": 0.2092127650976181,
      "learning_rate": 0.0004044132469112299,
      "loss": 2.0575,
      "step": 1253
    },
    {
      "epoch": 0.5749656121045392,
      "grad_norm": 0.23115339875221252,
      "learning_rate": 0.00040368435737307733,
      "loss": 2.1203,
      "step": 1254
    },
    {
      "epoch": 0.5754241173773499,
      "grad_norm": 0.20528317987918854,
      "learning_rate": 0.00040295568034315224,
      "loss": 1.9973,
      "step": 1255
    },
    {
      "epoch": 0.5758826226501604,
      "grad_norm": 0.20125994086265564,
      "learning_rate": 0.0004022272174291878,
      "loss": 2.083,
      "step": 1256
    },
    {
      "epoch": 0.5763411279229711,
      "grad_norm": 0.21145600080490112,
      "learning_rate": 0.0004014989702384449,
      "loss": 2.0055,
      "step": 1257
    },
    {
      "epoch": 0.5767996331957818,
      "grad_norm": 0.20329488813877106,
      "learning_rate": 0.00040077094037770843,
      "loss": 2.0308,
      "step": 1258
    },
    {
      "epoch": 0.5772581384685924,
      "grad_norm": 0.2436411827802658,
      "learning_rate": 0.0004000431294532838,
      "loss": 2.0364,
      "step": 1259
    },
    {
      "epoch": 0.5777166437414031,
      "grad_norm": 0.20469537377357483,
      "learning_rate": 0.0003993155390709935,
      "loss": 2.0411,
      "step": 1260
    },
    {
      "epoch": 0.5781751490142136,
      "grad_norm": 0.20830221474170685,
      "learning_rate": 0.0003985881708361729,
      "loss": 2.0961,
      "step": 1261
    },
    {
      "epoch": 0.5786336542870243,
      "grad_norm": 0.20512691140174866,
      "learning_rate": 0.00039786102635366784,
      "loss": 2.0127,
      "step": 1262
    },
    {
      "epoch": 0.579092159559835,
      "grad_norm": 0.21833814680576324,
      "learning_rate": 0.0003971341072278302,
      "loss": 2.0906,
      "step": 1263
    },
    {
      "epoch": 0.5795506648326456,
      "grad_norm": 0.20721854269504547,
      "learning_rate": 0.00039640741506251457,
      "loss": 2.1065,
      "step": 1264
    },
    {
      "epoch": 0.5800091701054562,
      "grad_norm": 0.2183152437210083,
      "learning_rate": 0.00039568095146107495,
      "loss": 2.0122,
      "step": 1265
    },
    {
      "epoch": 0.5804676753782668,
      "grad_norm": 0.2157343477010727,
      "learning_rate": 0.00039495471802636096,
      "loss": 2.1633,
      "step": 1266
    },
    {
      "epoch": 0.5809261806510775,
      "grad_norm": 0.21089483797550201,
      "learning_rate": 0.0003942287163607145,
      "loss": 1.9836,
      "step": 1267
    },
    {
      "epoch": 0.5813846859238881,
      "grad_norm": 0.2000267207622528,
      "learning_rate": 0.0003935029480659658,
      "loss": 2.0528,
      "step": 1268
    },
    {
      "epoch": 0.5818431911966988,
      "grad_norm": 0.20194357633590698,
      "learning_rate": 0.00039277741474343054,
      "loss": 2.11,
      "step": 1269
    },
    {
      "epoch": 0.5823016964695094,
      "grad_norm": 0.2069910317659378,
      "learning_rate": 0.0003920521179939057,
      "loss": 2.0584,
      "step": 1270
    },
    {
      "epoch": 0.58276020174232,
      "grad_norm": 0.19780677556991577,
      "learning_rate": 0.00039132705941766644,
      "loss": 2.1039,
      "step": 1271
    },
    {
      "epoch": 0.5832187070151307,
      "grad_norm": 0.20058110356330872,
      "learning_rate": 0.0003906022406144624,
      "loss": 1.9793,
      "step": 1272
    },
    {
      "epoch": 0.5836772122879413,
      "grad_norm": 0.20244082808494568,
      "learning_rate": 0.0003898776631835143,
      "loss": 2.0908,
      "step": 1273
    },
    {
      "epoch": 0.584135717560752,
      "grad_norm": 0.21247979998588562,
      "learning_rate": 0.00038915332872350994,
      "loss": 1.9771,
      "step": 1274
    },
    {
      "epoch": 0.5845942228335625,
      "grad_norm": 0.2014329433441162,
      "learning_rate": 0.00038842923883260135,
      "loss": 2.0083,
      "step": 1275
    },
    {
      "epoch": 0.5850527281063732,
      "grad_norm": 0.21621273458003998,
      "learning_rate": 0.00038770539510840093,
      "loss": 2.0778,
      "step": 1276
    },
    {
      "epoch": 0.5855112333791839,
      "grad_norm": 0.21339446306228638,
      "learning_rate": 0.00038698179914797783,
      "loss": 2.0714,
      "step": 1277
    },
    {
      "epoch": 0.5859697386519945,
      "grad_norm": 0.2146151065826416,
      "learning_rate": 0.0003862584525478545,
      "loss": 2.1025,
      "step": 1278
    },
    {
      "epoch": 0.5864282439248051,
      "grad_norm": 0.220952570438385,
      "learning_rate": 0.00038553535690400353,
      "loss": 2.1155,
      "step": 1279
    },
    {
      "epoch": 0.5868867491976157,
      "grad_norm": 0.20342887938022614,
      "learning_rate": 0.00038481251381184355,
      "loss": 2.0077,
      "step": 1280
    },
    {
      "epoch": 0.5873452544704264,
      "grad_norm": 0.20100677013397217,
      "learning_rate": 0.00038408992486623584,
      "loss": 2.0613,
      "step": 1281
    },
    {
      "epoch": 0.5878037597432371,
      "grad_norm": 0.2112731784582138,
      "learning_rate": 0.00038336759166148117,
      "loss": 2.1025,
      "step": 1282
    },
    {
      "epoch": 0.5882622650160477,
      "grad_norm": 0.21088185906410217,
      "learning_rate": 0.0003826455157913159,
      "loss": 1.9193,
      "step": 1283
    },
    {
      "epoch": 0.5887207702888583,
      "grad_norm": 0.21981248259544373,
      "learning_rate": 0.00038192369884890886,
      "loss": 2.0901,
      "step": 1284
    },
    {
      "epoch": 0.5891792755616689,
      "grad_norm": 0.20979179441928864,
      "learning_rate": 0.00038120214242685723,
      "loss": 2.0011,
      "step": 1285
    },
    {
      "epoch": 0.5896377808344796,
      "grad_norm": 0.2029990255832672,
      "learning_rate": 0.00038048084811718373,
      "loss": 1.9975,
      "step": 1286
    },
    {
      "epoch": 0.5900962861072903,
      "grad_norm": 0.20928402245044708,
      "learning_rate": 0.0003797598175113327,
      "loss": 2.0095,
      "step": 1287
    },
    {
      "epoch": 0.5905547913801009,
      "grad_norm": 0.19756776094436646,
      "learning_rate": 0.0003790390522001662,
      "loss": 1.9733,
      "step": 1288
    },
    {
      "epoch": 0.5910132966529115,
      "grad_norm": 0.21532626450061798,
      "learning_rate": 0.0003783185537739615,
      "loss": 2.0807,
      "step": 1289
    },
    {
      "epoch": 0.5914718019257221,
      "grad_norm": 0.22158078849315643,
      "learning_rate": 0.00037759832382240697,
      "loss": 2.0507,
      "step": 1290
    },
    {
      "epoch": 0.5919303071985328,
      "grad_norm": 0.21580110490322113,
      "learning_rate": 0.00037687836393459826,
      "loss": 2.0352,
      "step": 1291
    },
    {
      "epoch": 0.5923888124713435,
      "grad_norm": 0.21167203783988953,
      "learning_rate": 0.0003761586756990354,
      "loss": 2.0148,
      "step": 1292
    },
    {
      "epoch": 0.5928473177441541,
      "grad_norm": 0.20317263901233673,
      "learning_rate": 0.0003754392607036191,
      "loss": 2.0551,
      "step": 1293
    },
    {
      "epoch": 0.5933058230169647,
      "grad_norm": 0.2057163268327713,
      "learning_rate": 0.0003747201205356472,
      "loss": 2.0597,
      "step": 1294
    },
    {
      "epoch": 0.5937643282897753,
      "grad_norm": 0.20486022531986237,
      "learning_rate": 0.0003740012567818111,
      "loss": 2.1413,
      "step": 1295
    },
    {
      "epoch": 0.594222833562586,
      "grad_norm": 0.22216954827308655,
      "learning_rate": 0.0003732826710281922,
      "loss": 2.0111,
      "step": 1296
    },
    {
      "epoch": 0.5946813388353966,
      "grad_norm": 0.21419011056423187,
      "learning_rate": 0.0003725643648602588,
      "loss": 2.0139,
      "step": 1297
    },
    {
      "epoch": 0.5951398441082072,
      "grad_norm": 0.20638716220855713,
      "learning_rate": 0.0003718463398628621,
      "loss": 2.0463,
      "step": 1298
    },
    {
      "epoch": 0.5955983493810179,
      "grad_norm": 0.20625033974647522,
      "learning_rate": 0.0003711285976202331,
      "loss": 2.0807,
      "step": 1299
    },
    {
      "epoch": 0.5960568546538285,
      "grad_norm": 0.211114302277565,
      "learning_rate": 0.0003704111397159787,
      "loss": 2.0265,
      "step": 1300
    },
    {
      "epoch": 0.5965153599266392,
      "grad_norm": 0.2013825625181198,
      "learning_rate": 0.0003696939677330788,
      "loss": 2.0626,
      "step": 1301
    },
    {
      "epoch": 0.5969738651994498,
      "grad_norm": 0.20166388154029846,
      "learning_rate": 0.00036897708325388213,
      "loss": 1.9812,
      "step": 1302
    },
    {
      "epoch": 0.5974323704722604,
      "grad_norm": 0.20991401374340057,
      "learning_rate": 0.0003682604878601034,
      "loss": 2.0156,
      "step": 1303
    },
    {
      "epoch": 0.597890875745071,
      "grad_norm": 0.20064547657966614,
      "learning_rate": 0.000367544183132819,
      "loss": 1.9621,
      "step": 1304
    },
    {
      "epoch": 0.5983493810178817,
      "grad_norm": 0.2028827667236328,
      "learning_rate": 0.0003668281706524645,
      "loss": 1.9534,
      "step": 1305
    },
    {
      "epoch": 0.5988078862906924,
      "grad_norm": 0.21151524782180786,
      "learning_rate": 0.0003661124519988304,
      "loss": 2.0479,
      "step": 1306
    },
    {
      "epoch": 0.599266391563503,
      "grad_norm": 0.20141522586345673,
      "learning_rate": 0.00036539702875105893,
      "loss": 2.0345,
      "step": 1307
    },
    {
      "epoch": 0.5997248968363136,
      "grad_norm": 0.21383310854434967,
      "learning_rate": 0.0003646819024876406,
      "loss": 2.0609,
      "step": 1308
    },
    {
      "epoch": 0.6001834021091242,
      "grad_norm": 0.23201632499694824,
      "learning_rate": 0.0003639670747864107,
      "loss": 2.1431,
      "step": 1309
    },
    {
      "epoch": 0.6006419073819349,
      "grad_norm": 0.20978723466396332,
      "learning_rate": 0.00036325254722454584,
      "loss": 2.0628,
      "step": 1310
    },
    {
      "epoch": 0.6011004126547456,
      "grad_norm": 0.24303250014781952,
      "learning_rate": 0.00036253832137855997,
      "loss": 2.0691,
      "step": 1311
    },
    {
      "epoch": 0.6015589179275561,
      "grad_norm": 0.21711789071559906,
      "learning_rate": 0.00036182439882430183,
      "loss": 2.0669,
      "step": 1312
    },
    {
      "epoch": 0.6020174232003668,
      "grad_norm": 0.19928015768527985,
      "learning_rate": 0.00036111078113695096,
      "loss": 1.9693,
      "step": 1313
    },
    {
      "epoch": 0.6024759284731774,
      "grad_norm": 0.20862257480621338,
      "learning_rate": 0.0003603974698910139,
      "loss": 1.9747,
      "step": 1314
    },
    {
      "epoch": 0.6029344337459881,
      "grad_norm": 0.21825814247131348,
      "learning_rate": 0.0003596844666603214,
      "loss": 1.9963,
      "step": 1315
    },
    {
      "epoch": 0.6033929390187988,
      "grad_norm": 0.2057334929704666,
      "learning_rate": 0.0003589717730180245,
      "loss": 2.0489,
      "step": 1316
    },
    {
      "epoch": 0.6038514442916093,
      "grad_norm": 0.20310281217098236,
      "learning_rate": 0.00035825939053659117,
      "loss": 2.0815,
      "step": 1317
    },
    {
      "epoch": 0.60430994956442,
      "grad_norm": 0.20953218638896942,
      "learning_rate": 0.00035754732078780273,
      "loss": 1.977,
      "step": 1318
    },
    {
      "epoch": 0.6047684548372306,
      "grad_norm": 0.21018101274967194,
      "learning_rate": 0.00035683556534275076,
      "loss": 2.1343,
      "step": 1319
    },
    {
      "epoch": 0.6052269601100413,
      "grad_norm": 0.20807678997516632,
      "learning_rate": 0.00035612412577183303,
      "loss": 2.0181,
      "step": 1320
    },
    {
      "epoch": 0.605685465382852,
      "grad_norm": 0.20910735428333282,
      "learning_rate": 0.00035541300364475063,
      "loss": 1.9943,
      "step": 1321
    },
    {
      "epoch": 0.6061439706556625,
      "grad_norm": 0.1975119411945343,
      "learning_rate": 0.0003547022005305043,
      "loss": 2.0988,
      "step": 1322
    },
    {
      "epoch": 0.6066024759284732,
      "grad_norm": 0.20400920510292053,
      "learning_rate": 0.0003539917179973907,
      "loss": 2.0955,
      "step": 1323
    },
    {
      "epoch": 0.6070609812012838,
      "grad_norm": 0.1877986043691635,
      "learning_rate": 0.00035328155761299917,
      "loss": 2.0511,
      "step": 1324
    },
    {
      "epoch": 0.6075194864740945,
      "grad_norm": 0.2122442126274109,
      "learning_rate": 0.0003525717209442085,
      "loss": 1.9709,
      "step": 1325
    },
    {
      "epoch": 0.6079779917469051,
      "grad_norm": 0.19437754154205322,
      "learning_rate": 0.00035186220955718306,
      "loss": 1.8814,
      "step": 1326
    },
    {
      "epoch": 0.6084364970197157,
      "grad_norm": 0.21108679473400116,
      "learning_rate": 0.0003511530250173696,
      "loss": 2.05,
      "step": 1327
    },
    {
      "epoch": 0.6088950022925264,
      "grad_norm": 0.21027928590774536,
      "learning_rate": 0.00035044416888949364,
      "loss": 2.0172,
      "step": 1328
    },
    {
      "epoch": 0.609353507565337,
      "grad_norm": 0.21068333089351654,
      "learning_rate": 0.0003497356427375562,
      "loss": 2.0454,
      "step": 1329
    },
    {
      "epoch": 0.6098120128381477,
      "grad_norm": 0.2145102471113205,
      "learning_rate": 0.00034902744812483034,
      "loss": 2.0716,
      "step": 1330
    },
    {
      "epoch": 0.6102705181109582,
      "grad_norm": 0.21063804626464844,
      "learning_rate": 0.00034831958661385714,
      "loss": 2.0482,
      "step": 1331
    },
    {
      "epoch": 0.6107290233837689,
      "grad_norm": 0.20655828714370728,
      "learning_rate": 0.0003476120597664434,
      "loss": 2.0335,
      "step": 1332
    },
    {
      "epoch": 0.6111875286565795,
      "grad_norm": 0.21034331619739532,
      "learning_rate": 0.00034690486914365704,
      "loss": 2.0593,
      "step": 1333
    },
    {
      "epoch": 0.6116460339293902,
      "grad_norm": 0.21120253205299377,
      "learning_rate": 0.00034619801630582435,
      "loss": 2.1064,
      "step": 1334
    },
    {
      "epoch": 0.6121045392022009,
      "grad_norm": 0.2111034393310547,
      "learning_rate": 0.00034549150281252633,
      "loss": 2.002,
      "step": 1335
    },
    {
      "epoch": 0.6125630444750114,
      "grad_norm": 0.20359490811824799,
      "learning_rate": 0.0003447853302225952,
      "loss": 2.0232,
      "step": 1336
    },
    {
      "epoch": 0.6130215497478221,
      "grad_norm": 0.2161046862602234,
      "learning_rate": 0.00034407950009411126,
      "loss": 1.9407,
      "step": 1337
    },
    {
      "epoch": 0.6134800550206327,
      "grad_norm": 0.21669523417949677,
      "learning_rate": 0.00034337401398439873,
      "loss": 2.0029,
      "step": 1338
    },
    {
      "epoch": 0.6139385602934434,
      "grad_norm": 0.21236537396907806,
      "learning_rate": 0.00034266887345002305,
      "loss": 1.9665,
      "step": 1339
    },
    {
      "epoch": 0.6143970655662541,
      "grad_norm": 0.20678021013736725,
      "learning_rate": 0.0003419640800467874,
      "loss": 2.0651,
      "step": 1340
    },
    {
      "epoch": 0.6148555708390646,
      "grad_norm": 0.22239884734153748,
      "learning_rate": 0.0003412596353297288,
      "loss": 2.0634,
      "step": 1341
    },
    {
      "epoch": 0.6153140761118753,
      "grad_norm": 0.21979942917823792,
      "learning_rate": 0.00034055554085311493,
      "loss": 2.0232,
      "step": 1342
    },
    {
      "epoch": 0.6157725813846859,
      "grad_norm": 0.21928592026233673,
      "learning_rate": 0.00033985179817044105,
      "loss": 2.1216,
      "step": 1343
    },
    {
      "epoch": 0.6162310866574966,
      "grad_norm": 0.21501919627189636,
      "learning_rate": 0.0003391484088344257,
      "loss": 2.028,
      "step": 1344
    },
    {
      "epoch": 0.6166895919303071,
      "grad_norm": 0.2109012007713318,
      "learning_rate": 0.00033844537439700807,
      "loss": 2.1485,
      "step": 1345
    },
    {
      "epoch": 0.6171480972031178,
      "grad_norm": 0.2040739506483078,
      "learning_rate": 0.00033774269640934445,
      "loss": 2.0816,
      "step": 1346
    },
    {
      "epoch": 0.6176066024759285,
      "grad_norm": 0.19894567131996155,
      "learning_rate": 0.0003370403764218045,
      "loss": 1.991,
      "step": 1347
    },
    {
      "epoch": 0.6180651077487391,
      "grad_norm": 0.2011205404996872,
      "learning_rate": 0.000336338415983968,
      "loss": 1.9793,
      "step": 1348
    },
    {
      "epoch": 0.6185236130215498,
      "grad_norm": 0.19460994005203247,
      "learning_rate": 0.00033563681664462155,
      "loss": 1.9851,
      "step": 1349
    },
    {
      "epoch": 0.6189821182943603,
      "grad_norm": 0.21164923906326294,
      "learning_rate": 0.000334935579951755,
      "loss": 2.1542,
      "step": 1350
    },
    {
      "epoch": 0.619440623567171,
      "grad_norm": 0.20313669741153717,
      "learning_rate": 0.0003342347074525578,
      "loss": 2.0025,
      "step": 1351
    },
    {
      "epoch": 0.6198991288399817,
      "grad_norm": 0.21033664047718048,
      "learning_rate": 0.0003335342006934161,
      "loss": 2.0927,
      "step": 1352
    },
    {
      "epoch": 0.6203576341127923,
      "grad_norm": 0.20695436000823975,
      "learning_rate": 0.00033283406121990914,
      "loss": 2.0661,
      "step": 1353
    },
    {
      "epoch": 0.620816139385603,
      "grad_norm": 0.2113455981016159,
      "learning_rate": 0.0003321342905768057,
      "loss": 1.9949,
      "step": 1354
    },
    {
      "epoch": 0.6212746446584135,
      "grad_norm": 0.21170936524868011,
      "learning_rate": 0.00033143489030806086,
      "loss": 2.0198,
      "step": 1355
    },
    {
      "epoch": 0.6217331499312242,
      "grad_norm": 0.208674818277359,
      "learning_rate": 0.00033073586195681227,
      "loss": 1.9838,
      "step": 1356
    },
    {
      "epoch": 0.6221916552040349,
      "grad_norm": 0.21607069671154022,
      "learning_rate": 0.00033003720706537736,
      "loss": 1.9444,
      "step": 1357
    },
    {
      "epoch": 0.6226501604768455,
      "grad_norm": 0.21917787194252014,
      "learning_rate": 0.0003293389271752492,
      "loss": 2.0214,
      "step": 1358
    },
    {
      "epoch": 0.6231086657496562,
      "grad_norm": 0.21066710352897644,
      "learning_rate": 0.00032864102382709374,
      "loss": 2.1414,
      "step": 1359
    },
    {
      "epoch": 0.6235671710224667,
      "grad_norm": 0.21104732155799866,
      "learning_rate": 0.000327943498560746,
      "loss": 2.0592,
      "step": 1360
    },
    {
      "epoch": 0.6240256762952774,
      "grad_norm": 0.2087271809577942,
      "learning_rate": 0.00032724635291520694,
      "loss": 2.0769,
      "step": 1361
    },
    {
      "epoch": 0.624484181568088,
      "grad_norm": 0.21306948363780975,
      "learning_rate": 0.00032654958842863967,
      "loss": 2.0295,
      "step": 1362
    },
    {
      "epoch": 0.6249426868408987,
      "grad_norm": 0.20913147926330566,
      "learning_rate": 0.0003258532066383667,
      "loss": 1.9524,
      "step": 1363
    },
    {
      "epoch": 0.6254011921137093,
      "grad_norm": 0.19989560544490814,
      "learning_rate": 0.000325157209080866,
      "loss": 1.967,
      "step": 1364
    },
    {
      "epoch": 0.6258596973865199,
      "grad_norm": 0.20310388505458832,
      "learning_rate": 0.00032446159729176743,
      "loss": 2.0749,
      "step": 1365
    },
    {
      "epoch": 0.6263182026593306,
      "grad_norm": 0.21072514355182648,
      "learning_rate": 0.0003237663728058502,
      "loss": 1.9644,
      "step": 1366
    },
    {
      "epoch": 0.6267767079321412,
      "grad_norm": 0.21206192672252655,
      "learning_rate": 0.0003230715371570389,
      "loss": 2.0864,
      "step": 1367
    },
    {
      "epoch": 0.6272352132049519,
      "grad_norm": 0.20161403715610504,
      "learning_rate": 0.00032237709187839996,
      "loss": 1.9598,
      "step": 1368
    },
    {
      "epoch": 0.6276937184777625,
      "grad_norm": 0.19552923738956451,
      "learning_rate": 0.0003216830385021388,
      "loss": 1.9535,
      "step": 1369
    },
    {
      "epoch": 0.6281522237505731,
      "grad_norm": 0.2102045863866806,
      "learning_rate": 0.0003209893785595959,
      "loss": 2.1435,
      "step": 1370
    },
    {
      "epoch": 0.6286107290233838,
      "grad_norm": 0.20964489877223969,
      "learning_rate": 0.00032029611358124366,
      "loss": 1.9755,
      "step": 1371
    },
    {
      "epoch": 0.6290692342961944,
      "grad_norm": 0.23071300983428955,
      "learning_rate": 0.00031960324509668336,
      "loss": 2.0198,
      "step": 1372
    },
    {
      "epoch": 0.6295277395690051,
      "grad_norm": 0.2121468037366867,
      "learning_rate": 0.0003189107746346412,
      "loss": 2.0739,
      "step": 1373
    },
    {
      "epoch": 0.6299862448418156,
      "grad_norm": 0.2064194679260254,
      "learning_rate": 0.0003182187037229653,
      "loss": 2.0087,
      "step": 1374
    },
    {
      "epoch": 0.6304447501146263,
      "grad_norm": 0.207213893532753,
      "learning_rate": 0.0003175270338886221,
      "loss": 2.0926,
      "step": 1375
    },
    {
      "epoch": 0.630903255387437,
      "grad_norm": 0.21379680931568146,
      "learning_rate": 0.00031683576665769345,
      "loss": 2.0319,
      "step": 1376
    },
    {
      "epoch": 0.6313617606602476,
      "grad_norm": 0.20832125842571259,
      "learning_rate": 0.0003161449035553724,
      "loss": 1.9686,
      "step": 1377
    },
    {
      "epoch": 0.6318202659330582,
      "grad_norm": 0.19952528178691864,
      "learning_rate": 0.00031545444610596077,
      "loss": 2.0675,
      "step": 1378
    },
    {
      "epoch": 0.6322787712058688,
      "grad_norm": 0.20252546668052673,
      "learning_rate": 0.000314764395832865,
      "loss": 2.0218,
      "step": 1379
    },
    {
      "epoch": 0.6327372764786795,
      "grad_norm": 0.2084491103887558,
      "learning_rate": 0.0003140747542585934,
      "loss": 1.9667,
      "step": 1380
    },
    {
      "epoch": 0.6331957817514902,
      "grad_norm": 0.20819610357284546,
      "learning_rate": 0.00031338552290475266,
      "loss": 2.0461,
      "step": 1381
    },
    {
      "epoch": 0.6336542870243008,
      "grad_norm": 0.20372039079666138,
      "learning_rate": 0.00031269670329204396,
      "loss": 1.9687,
      "step": 1382
    },
    {
      "epoch": 0.6341127922971114,
      "grad_norm": 0.2117212861776352,
      "learning_rate": 0.0003120082969402604,
      "loss": 2.0809,
      "step": 1383
    },
    {
      "epoch": 0.634571297569922,
      "grad_norm": 0.20521347224712372,
      "learning_rate": 0.00031132030536828314,
      "loss": 2.0962,
      "step": 1384
    },
    {
      "epoch": 0.6350298028427327,
      "grad_norm": 0.2173711657524109,
      "learning_rate": 0.00031063273009407805,
      "loss": 2.0043,
      "step": 1385
    },
    {
      "epoch": 0.6354883081155434,
      "grad_norm": 0.2165997326374054,
      "learning_rate": 0.00030994557263469265,
      "loss": 2.0459,
      "step": 1386
    },
    {
      "epoch": 0.635946813388354,
      "grad_norm": 0.2022165209054947,
      "learning_rate": 0.0003092588345062526,
      "loss": 2.0211,
      "step": 1387
    },
    {
      "epoch": 0.6364053186611646,
      "grad_norm": 0.2137582153081894,
      "learning_rate": 0.0003085725172239582,
      "loss": 2.0867,
      "step": 1388
    },
    {
      "epoch": 0.6368638239339752,
      "grad_norm": 0.20703867077827454,
      "learning_rate": 0.0003078866223020815,
      "loss": 2.0328,
      "step": 1389
    },
    {
      "epoch": 0.6373223292067859,
      "grad_norm": 0.20792417228221893,
      "learning_rate": 0.0003072011512539624,
      "loss": 1.9992,
      "step": 1390
    },
    {
      "epoch": 0.6377808344795965,
      "grad_norm": 0.21017327904701233,
      "learning_rate": 0.00030651610559200574,
      "loss": 2.0762,
      "step": 1391
    },
    {
      "epoch": 0.6382393397524071,
      "grad_norm": 0.21039779484272003,
      "learning_rate": 0.00030583148682767757,
      "loss": 2.0352,
      "step": 1392
    },
    {
      "epoch": 0.6386978450252178,
      "grad_norm": 0.20955179631710052,
      "learning_rate": 0.00030514729647150243,
      "loss": 2.017,
      "step": 1393
    },
    {
      "epoch": 0.6391563502980284,
      "grad_norm": 0.2080843448638916,
      "learning_rate": 0.0003044635360330592,
      "loss": 1.9784,
      "step": 1394
    },
    {
      "epoch": 0.6396148555708391,
      "grad_norm": 0.21270842850208282,
      "learning_rate": 0.00030378020702097845,
      "loss": 2.0362,
      "step": 1395
    },
    {
      "epoch": 0.6400733608436497,
      "grad_norm": 0.21405519545078278,
      "learning_rate": 0.000303097310942939,
      "loss": 2.0778,
      "step": 1396
    },
    {
      "epoch": 0.6405318661164603,
      "grad_norm": 0.20308026671409607,
      "learning_rate": 0.0003024148493056641,
      "loss": 2.0097,
      "step": 1397
    },
    {
      "epoch": 0.640990371389271,
      "grad_norm": 0.20590254664421082,
      "learning_rate": 0.00030173282361491865,
      "loss": 1.9541,
      "step": 1398
    },
    {
      "epoch": 0.6414488766620816,
      "grad_norm": 0.21124644577503204,
      "learning_rate": 0.0003010512353755057,
      "loss": 1.9681,
      "step": 1399
    },
    {
      "epoch": 0.6419073819348923,
      "grad_norm": 0.20847435295581818,
      "learning_rate": 0.00030037008609126313,
      "loss": 2.0093,
      "step": 1400
    },
    {
      "epoch": 0.6423658872077029,
      "grad_norm": 0.20726299285888672,
      "learning_rate": 0.0002996893772650602,
      "loss": 1.9669,
      "step": 1401
    },
    {
      "epoch": 0.6428243924805135,
      "grad_norm": 0.20740748941898346,
      "learning_rate": 0.0002990091103987945,
      "loss": 2.1185,
      "step": 1402
    },
    {
      "epoch": 0.6432828977533241,
      "grad_norm": 0.21779777109622955,
      "learning_rate": 0.0002983292869933886,
      "loss": 2.1085,
      "step": 1403
    },
    {
      "epoch": 0.6437414030261348,
      "grad_norm": 0.20268970727920532,
      "learning_rate": 0.0002976499085487862,
      "loss": 2.0349,
      "step": 1404
    },
    {
      "epoch": 0.6441999082989455,
      "grad_norm": 0.2069568932056427,
      "learning_rate": 0.00029697097656394963,
      "loss": 2.0242,
      "step": 1405
    },
    {
      "epoch": 0.6446584135717561,
      "grad_norm": 0.20344099402427673,
      "learning_rate": 0.00029629249253685595,
      "loss": 1.9456,
      "step": 1406
    },
    {
      "epoch": 0.6451169188445667,
      "grad_norm": 0.20613251626491547,
      "learning_rate": 0.00029561445796449416,
      "loss": 2.112,
      "step": 1407
    },
    {
      "epoch": 0.6455754241173773,
      "grad_norm": 0.2087947577238083,
      "learning_rate": 0.0002949368743428612,
      "loss": 1.999,
      "step": 1408
    },
    {
      "epoch": 0.646033929390188,
      "grad_norm": 0.20399583876132965,
      "learning_rate": 0.0002942597431669593,
      "loss": 2.0094,
      "step": 1409
    },
    {
      "epoch": 0.6464924346629987,
      "grad_norm": 0.21521978080272675,
      "learning_rate": 0.0002935830659307924,
      "loss": 2.0553,
      "step": 1410
    },
    {
      "epoch": 0.6469509399358092,
      "grad_norm": 0.21704481542110443,
      "learning_rate": 0.0002929068441273629,
      "loss": 2.0207,
      "step": 1411
    },
    {
      "epoch": 0.6474094452086199,
      "grad_norm": 0.21603667736053467,
      "learning_rate": 0.0002922310792486681,
      "loss": 2.0207,
      "step": 1412
    },
    {
      "epoch": 0.6478679504814305,
      "grad_norm": 0.20676253736019135,
      "learning_rate": 0.00029155577278569745,
      "loss": 1.9846,
      "step": 1413
    },
    {
      "epoch": 0.6483264557542412,
      "grad_norm": 0.21380478143692017,
      "learning_rate": 0.00029088092622842895,
      "loss": 2.109,
      "step": 1414
    },
    {
      "epoch": 0.6487849610270519,
      "grad_norm": 0.200730100274086,
      "learning_rate": 0.00029020654106582544,
      "loss": 2.0711,
      "step": 1415
    },
    {
      "epoch": 0.6492434662998624,
      "grad_norm": 0.21288162469863892,
      "learning_rate": 0.0002895326187858326,
      "loss": 2.021,
      "step": 1416
    },
    {
      "epoch": 0.6497019715726731,
      "grad_norm": 0.21866631507873535,
      "learning_rate": 0.00028885916087537377,
      "loss": 2.0634,
      "step": 1417
    },
    {
      "epoch": 0.6501604768454837,
      "grad_norm": 0.21763920783996582,
      "learning_rate": 0.00028818616882034877,
      "loss": 2.0657,
      "step": 1418
    },
    {
      "epoch": 0.6506189821182944,
      "grad_norm": 0.21188122034072876,
      "learning_rate": 0.0002875136441056286,
      "loss": 1.9732,
      "step": 1419
    },
    {
      "epoch": 0.651077487391105,
      "grad_norm": 0.22483225166797638,
      "learning_rate": 0.000286841588215054,
      "loss": 2.0917,
      "step": 1420
    },
    {
      "epoch": 0.6515359926639156,
      "grad_norm": 0.21220411360263824,
      "learning_rate": 0.0002861700026314308,
      "loss": 2.0313,
      "step": 1421
    },
    {
      "epoch": 0.6519944979367263,
      "grad_norm": 0.21440936625003815,
      "learning_rate": 0.00028549888883652686,
      "loss": 1.9677,
      "step": 1422
    },
    {
      "epoch": 0.6524530032095369,
      "grad_norm": 0.2158776819705963,
      "learning_rate": 0.00028482824831107,
      "loss": 2.085,
      "step": 1423
    },
    {
      "epoch": 0.6529115084823476,
      "grad_norm": 0.21344362199306488,
      "learning_rate": 0.000284158082534743,
      "loss": 1.9525,
      "step": 1424
    },
    {
      "epoch": 0.6533700137551581,
      "grad_norm": 0.2241325080394745,
      "learning_rate": 0.00028348839298618177,
      "loss": 1.9835,
      "step": 1425
    },
    {
      "epoch": 0.6538285190279688,
      "grad_norm": 0.20661838352680206,
      "learning_rate": 0.0002828191811429709,
      "loss": 2.008,
      "step": 1426
    },
    {
      "epoch": 0.6542870243007795,
      "grad_norm": 0.21329088509082794,
      "learning_rate": 0.00028215044848164164,
      "loss": 2.0621,
      "step": 1427
    },
    {
      "epoch": 0.6547455295735901,
      "grad_norm": 0.21919915080070496,
      "learning_rate": 0.00028148219647766747,
      "loss": 2.1529,
      "step": 1428
    },
    {
      "epoch": 0.6552040348464008,
      "grad_norm": 0.21397031843662262,
      "learning_rate": 0.00028081442660546124,
      "loss": 1.978,
      "step": 1429
    },
    {
      "epoch": 0.6556625401192113,
      "grad_norm": 0.22627611458301544,
      "learning_rate": 0.0002801471403383728,
      "loss": 1.9626,
      "step": 1430
    },
    {
      "epoch": 0.656121045392022,
      "grad_norm": 0.2046988159418106,
      "learning_rate": 0.00027948033914868415,
      "loss": 2.0171,
      "step": 1431
    },
    {
      "epoch": 0.6565795506648326,
      "grad_norm": 0.21059556305408478,
      "learning_rate": 0.00027881402450760775,
      "loss": 2.0691,
      "step": 1432
    },
    {
      "epoch": 0.6570380559376433,
      "grad_norm": 0.20770058035850525,
      "learning_rate": 0.00027814819788528165,
      "loss": 2.0745,
      "step": 1433
    },
    {
      "epoch": 0.657496561210454,
      "grad_norm": 0.21462233364582062,
      "learning_rate": 0.00027748286075076836,
      "loss": 1.9846,
      "step": 1434
    },
    {
      "epoch": 0.6579550664832645,
      "grad_norm": 0.2109721302986145,
      "learning_rate": 0.00027681801457204937,
      "loss": 1.9776,
      "step": 1435
    },
    {
      "epoch": 0.6584135717560752,
      "grad_norm": 0.20669925212860107,
      "learning_rate": 0.00027615366081602306,
      "loss": 2.1032,
      "step": 1436
    },
    {
      "epoch": 0.6588720770288858,
      "grad_norm": 0.21343183517456055,
      "learning_rate": 0.0002754898009485021,
      "loss": 1.977,
      "step": 1437
    },
    {
      "epoch": 0.6593305823016965,
      "grad_norm": 0.20209427177906036,
      "learning_rate": 0.0002748264364342085,
      "loss": 1.9953,
      "step": 1438
    },
    {
      "epoch": 0.6597890875745072,
      "grad_norm": 0.21890811622142792,
      "learning_rate": 0.00027416356873677204,
      "loss": 1.9811,
      "step": 1439
    },
    {
      "epoch": 0.6602475928473177,
      "grad_norm": 0.21359935402870178,
      "learning_rate": 0.0002735011993187258,
      "loss": 1.9957,
      "step": 1440
    },
    {
      "epoch": 0.6607060981201284,
      "grad_norm": 0.21052290499210358,
      "learning_rate": 0.0002728393296415042,
      "loss": 2.1318,
      "step": 1441
    },
    {
      "epoch": 0.661164603392939,
      "grad_norm": 0.20989243686199188,
      "learning_rate": 0.00027217796116543817,
      "loss": 2.0373,
      "step": 1442
    },
    {
      "epoch": 0.6616231086657497,
      "grad_norm": 0.20901499688625336,
      "learning_rate": 0.0002715170953497532,
      "loss": 1.9458,
      "step": 1443
    },
    {
      "epoch": 0.6620816139385602,
      "grad_norm": 0.2207915186882019,
      "learning_rate": 0.00027085673365256614,
      "loss": 1.9999,
      "step": 1444
    },
    {
      "epoch": 0.6625401192113709,
      "grad_norm": 0.21182781457901,
      "learning_rate": 0.00027019687753088075,
      "loss": 2.0844,
      "step": 1445
    },
    {
      "epoch": 0.6629986244841816,
      "grad_norm": 0.20463135838508606,
      "learning_rate": 0.00026953752844058597,
      "loss": 2.0107,
      "step": 1446
    },
    {
      "epoch": 0.6634571297569922,
      "grad_norm": 0.20771057903766632,
      "learning_rate": 0.0002688786878364516,
      "loss": 2.0634,
      "step": 1447
    },
    {
      "epoch": 0.6639156350298029,
      "grad_norm": 0.20623765885829926,
      "learning_rate": 0.00026822035717212597,
      "loss": 2.0773,
      "step": 1448
    },
    {
      "epoch": 0.6643741403026134,
      "grad_norm": 0.20438608527183533,
      "learning_rate": 0.00026756253790013193,
      "loss": 1.9592,
      "step": 1449
    },
    {
      "epoch": 0.6648326455754241,
      "grad_norm": 0.20699217915534973,
      "learning_rate": 0.0002669052314718641,
      "loss": 1.985,
      "step": 1450
    },
    {
      "epoch": 0.6652911508482348,
      "grad_norm": 0.20409883558750153,
      "learning_rate": 0.0002662484393375855,
      "loss": 1.9532,
      "step": 1451
    },
    {
      "epoch": 0.6657496561210454,
      "grad_norm": 0.22611203789710999,
      "learning_rate": 0.00026559216294642446,
      "loss": 2.0178,
      "step": 1452
    },
    {
      "epoch": 0.6662081613938561,
      "grad_norm": 0.20953720808029175,
      "learning_rate": 0.0002649364037463718,
      "loss": 2.0708,
      "step": 1453
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.21088114380836487,
      "learning_rate": 0.0002642811631842764,
      "loss": 2.0454,
      "step": 1454
    },
    {
      "epoch": 0.6671251719394773,
      "grad_norm": 0.2060849666595459,
      "learning_rate": 0.0002636264427058439,
      "loss": 2.0114,
      "step": 1455
    },
    {
      "epoch": 0.667583677212288,
      "grad_norm": 0.21274107694625854,
      "learning_rate": 0.00026297224375563123,
      "loss": 2.0041,
      "step": 1456
    },
    {
      "epoch": 0.6680421824850986,
      "grad_norm": 0.21321925520896912,
      "learning_rate": 0.00026231856777704575,
      "loss": 2.0398,
      "step": 1457
    },
    {
      "epoch": 0.6685006877579092,
      "grad_norm": 0.21353711187839508,
      "learning_rate": 0.00026166541621234026,
      "loss": 2.001,
      "step": 1458
    },
    {
      "epoch": 0.6689591930307198,
      "grad_norm": 0.2161954939365387,
      "learning_rate": 0.00026101279050261045,
      "loss": 1.9163,
      "step": 1459
    },
    {
      "epoch": 0.6694176983035305,
      "grad_norm": 0.22429299354553223,
      "learning_rate": 0.00026036069208779247,
      "loss": 2.0734,
      "step": 1460
    },
    {
      "epoch": 0.6698762035763411,
      "grad_norm": 0.21448646485805511,
      "learning_rate": 0.0002597091224066581,
      "loss": 1.9824,
      "step": 1461
    },
    {
      "epoch": 0.6703347088491518,
      "grad_norm": 0.2162591814994812,
      "learning_rate": 0.00025905808289681365,
      "loss": 2.0204,
      "step": 1462
    },
    {
      "epoch": 0.6707932141219624,
      "grad_norm": 0.21993832290172577,
      "learning_rate": 0.0002584075749946946,
      "loss": 2.0312,
      "step": 1463
    },
    {
      "epoch": 0.671251719394773,
      "grad_norm": 0.21477247774600983,
      "learning_rate": 0.00025775760013556424,
      "loss": 2.0123,
      "step": 1464
    },
    {
      "epoch": 0.6717102246675837,
      "grad_norm": 0.2080913782119751,
      "learning_rate": 0.0002571081597535095,
      "loss": 2.0474,
      "step": 1465
    },
    {
      "epoch": 0.6721687299403943,
      "grad_norm": 0.2216774821281433,
      "learning_rate": 0.00025645925528143776,
      "loss": 2.0749,
      "step": 1466
    },
    {
      "epoch": 0.672627235213205,
      "grad_norm": 0.2379143089056015,
      "learning_rate": 0.0002558108881510747,
      "loss": 2.0751,
      "step": 1467
    },
    {
      "epoch": 0.6730857404860155,
      "grad_norm": 0.22319257259368896,
      "learning_rate": 0.00025516305979295963,
      "loss": 1.9571,
      "step": 1468
    },
    {
      "epoch": 0.6735442457588262,
      "grad_norm": 0.2063499093055725,
      "learning_rate": 0.0002545157716364439,
      "loss": 2.0515,
      "step": 1469
    },
    {
      "epoch": 0.6740027510316369,
      "grad_norm": 0.20929093658924103,
      "learning_rate": 0.00025386902510968624,
      "loss": 1.9571,
      "step": 1470
    },
    {
      "epoch": 0.6744612563044475,
      "grad_norm": 0.20345915853977203,
      "learning_rate": 0.00025322282163965095,
      "loss": 2.0938,
      "step": 1471
    },
    {
      "epoch": 0.6749197615772582,
      "grad_norm": 0.19785180687904358,
      "learning_rate": 0.00025257716265210384,
      "loss": 2.0202,
      "step": 1472
    },
    {
      "epoch": 0.6753782668500687,
      "grad_norm": 0.22299924492835999,
      "learning_rate": 0.0002519320495716091,
      "loss": 1.9538,
      "step": 1473
    },
    {
      "epoch": 0.6758367721228794,
      "grad_norm": 0.21873433887958527,
      "learning_rate": 0.00025128748382152716,
      "loss": 2.0715,
      "step": 1474
    },
    {
      "epoch": 0.6762952773956901,
      "grad_norm": 0.19794774055480957,
      "learning_rate": 0.00025064346682401016,
      "loss": 1.9909,
      "step": 1475
    },
    {
      "epoch": 0.6767537826685007,
      "grad_norm": 0.21572145819664001,
      "learning_rate": 0.0002500000000000001,
      "loss": 2.0355,
      "step": 1476
    },
    {
      "epoch": 0.6772122879413113,
      "grad_norm": 0.21858660876750946,
      "learning_rate": 0.0002493570847692246,
      "loss": 2.033,
      "step": 1477
    },
    {
      "epoch": 0.6776707932141219,
      "grad_norm": 0.2128642350435257,
      "learning_rate": 0.00024871472255019424,
      "loss": 1.9731,
      "step": 1478
    },
    {
      "epoch": 0.6781292984869326,
      "grad_norm": 0.2108616977930069,
      "learning_rate": 0.00024807291476019994,
      "loss": 2.0329,
      "step": 1479
    },
    {
      "epoch": 0.6785878037597433,
      "grad_norm": 0.20993788540363312,
      "learning_rate": 0.00024743166281530877,
      "loss": 1.9619,
      "step": 1480
    },
    {
      "epoch": 0.6790463090325539,
      "grad_norm": 0.21862910687923431,
      "learning_rate": 0.000246790968130362,
      "loss": 2.0593,
      "step": 1481
    },
    {
      "epoch": 0.6795048143053645,
      "grad_norm": 0.2194235622882843,
      "learning_rate": 0.0002461508321189706,
      "loss": 1.985,
      "step": 1482
    },
    {
      "epoch": 0.6799633195781751,
      "grad_norm": 0.20968922972679138,
      "learning_rate": 0.00024551125619351385,
      "loss": 2.0229,
      "step": 1483
    },
    {
      "epoch": 0.6804218248509858,
      "grad_norm": 0.21624493598937988,
      "learning_rate": 0.00024487224176513453,
      "loss": 2.007,
      "step": 1484
    },
    {
      "epoch": 0.6808803301237965,
      "grad_norm": 0.20427393913269043,
      "learning_rate": 0.0002442337902437365,
      "loss": 2.0756,
      "step": 1485
    },
    {
      "epoch": 0.6813388353966071,
      "grad_norm": 0.21779970824718475,
      "learning_rate": 0.0002435959030379824,
      "loss": 1.9994,
      "step": 1486
    },
    {
      "epoch": 0.6817973406694177,
      "grad_norm": 0.20751647651195526,
      "learning_rate": 0.00024295858155528888,
      "loss": 2.0203,
      "step": 1487
    },
    {
      "epoch": 0.6822558459422283,
      "grad_norm": 0.21935631334781647,
      "learning_rate": 0.00024232182720182523,
      "loss": 2.0449,
      "step": 1488
    },
    {
      "epoch": 0.682714351215039,
      "grad_norm": 0.22384728491306305,
      "learning_rate": 0.00024168564138250855,
      "loss": 2.0584,
      "step": 1489
    },
    {
      "epoch": 0.6831728564878496,
      "grad_norm": 0.22040314972400665,
      "learning_rate": 0.00024105002550100246,
      "loss": 2.0537,
      "step": 1490
    },
    {
      "epoch": 0.6836313617606602,
      "grad_norm": 0.2070879191160202,
      "learning_rate": 0.00024041498095971254,
      "loss": 2.0409,
      "step": 1491
    },
    {
      "epoch": 0.6840898670334709,
      "grad_norm": 0.21917015314102173,
      "learning_rate": 0.0002397805091597835,
      "loss": 2.0753,
      "step": 1492
    },
    {
      "epoch": 0.6845483723062815,
      "grad_norm": 0.22013387084007263,
      "learning_rate": 0.0002391466115010973,
      "loss": 2.0298,
      "step": 1493
    },
    {
      "epoch": 0.6850068775790922,
      "grad_norm": 0.2144373059272766,
      "learning_rate": 0.00023851328938226808,
      "loss": 2.0533,
      "step": 1494
    },
    {
      "epoch": 0.6854653828519028,
      "grad_norm": 0.2191096842288971,
      "learning_rate": 0.00023788054420064109,
      "loss": 1.9925,
      "step": 1495
    },
    {
      "epoch": 0.6859238881247134,
      "grad_norm": 0.2143266350030899,
      "learning_rate": 0.00023724837735228773,
      "loss": 1.9924,
      "step": 1496
    },
    {
      "epoch": 0.686382393397524,
      "grad_norm": 0.21332935988903046,
      "learning_rate": 0.00023661679023200422,
      "loss": 2.0473,
      "step": 1497
    },
    {
      "epoch": 0.6868408986703347,
      "grad_norm": 0.2160414159297943,
      "learning_rate": 0.00023598578423330714,
      "loss": 2.0051,
      "step": 1498
    },
    {
      "epoch": 0.6872994039431454,
      "grad_norm": 0.2102707177400589,
      "learning_rate": 0.00023535536074843083,
      "loss": 1.9918,
      "step": 1499
    },
    {
      "epoch": 0.687757909215956,
      "grad_norm": 0.21358685195446014,
      "learning_rate": 0.00023472552116832502,
      "loss": 2.0438,
      "step": 1500
    },
    {
      "epoch": 0.6882164144887666,
      "grad_norm": 0.2078513652086258,
      "learning_rate": 0.0002340962668826503,
      "loss": 2.047,
      "step": 1501
    },
    {
      "epoch": 0.6886749197615772,
      "grad_norm": 0.21196392178535461,
      "learning_rate": 0.00023346759927977663,
      "loss": 2.0714,
      "step": 1502
    },
    {
      "epoch": 0.6891334250343879,
      "grad_norm": 0.20410814881324768,
      "learning_rate": 0.0002328395197467789,
      "loss": 2.0662,
      "step": 1503
    },
    {
      "epoch": 0.6895919303071986,
      "grad_norm": 0.20888519287109375,
      "learning_rate": 0.00023221202966943515,
      "loss": 2.032,
      "step": 1504
    },
    {
      "epoch": 0.6900504355800092,
      "grad_norm": 0.21249207854270935,
      "learning_rate": 0.0002315851304322223,
      "loss": 2.033,
      "step": 1505
    },
    {
      "epoch": 0.6905089408528198,
      "grad_norm": 0.2116980254650116,
      "learning_rate": 0.0002309588234183137,
      "loss": 2.0871,
      "step": 1506
    },
    {
      "epoch": 0.6909674461256304,
      "grad_norm": 0.2137322872877121,
      "learning_rate": 0.00023033311000957653,
      "loss": 2.0048,
      "step": 1507
    },
    {
      "epoch": 0.6914259513984411,
      "grad_norm": 0.21266447007656097,
      "learning_rate": 0.00022970799158656758,
      "loss": 2.0268,
      "step": 1508
    },
    {
      "epoch": 0.6918844566712518,
      "grad_norm": 0.2163049578666687,
      "learning_rate": 0.0002290834695285316,
      "loss": 2.0295,
      "step": 1509
    },
    {
      "epoch": 0.6923429619440623,
      "grad_norm": 0.21484699845314026,
      "learning_rate": 0.00022845954521339678,
      "loss": 2.0884,
      "step": 1510
    },
    {
      "epoch": 0.692801467216873,
      "grad_norm": 0.20561583340168,
      "learning_rate": 0.0002278362200177732,
      "loss": 2.0403,
      "step": 1511
    },
    {
      "epoch": 0.6932599724896836,
      "grad_norm": 0.21210668981075287,
      "learning_rate": 0.00022721349531694852,
      "loss": 1.9899,
      "step": 1512
    },
    {
      "epoch": 0.6937184777624943,
      "grad_norm": 0.21584658324718475,
      "learning_rate": 0.0002265913724848855,
      "loss": 2.0211,
      "step": 1513
    },
    {
      "epoch": 0.694176983035305,
      "grad_norm": 0.2146245688199997,
      "learning_rate": 0.00022596985289421946,
      "loss": 2.0266,
      "step": 1514
    },
    {
      "epoch": 0.6946354883081155,
      "grad_norm": 0.21372953057289124,
      "learning_rate": 0.00022534893791625405,
      "loss": 1.9888,
      "step": 1515
    },
    {
      "epoch": 0.6950939935809262,
      "grad_norm": 0.21314692497253418,
      "learning_rate": 0.00022472862892095968,
      "loss": 2.0193,
      "step": 1516
    },
    {
      "epoch": 0.6955524988537368,
      "grad_norm": 0.2239692062139511,
      "learning_rate": 0.00022410892727696896,
      "loss": 2.1292,
      "step": 1517
    },
    {
      "epoch": 0.6960110041265475,
      "grad_norm": 0.20880535244941711,
      "learning_rate": 0.0002234898343515751,
      "loss": 2.0367,
      "step": 1518
    },
    {
      "epoch": 0.6964695093993581,
      "grad_norm": 0.20821888744831085,
      "learning_rate": 0.00022287135151072792,
      "loss": 2.0106,
      "step": 1519
    },
    {
      "epoch": 0.6969280146721687,
      "grad_norm": 0.20391957461833954,
      "learning_rate": 0.00022225348011903096,
      "loss": 2.0325,
      "step": 1520
    },
    {
      "epoch": 0.6973865199449794,
      "grad_norm": 0.21092291176319122,
      "learning_rate": 0.0002216362215397393,
      "loss": 1.9736,
      "step": 1521
    },
    {
      "epoch": 0.69784502521779,
      "grad_norm": 0.20765821635723114,
      "learning_rate": 0.00022101957713475522,
      "loss": 1.9897,
      "step": 1522
    },
    {
      "epoch": 0.6983035304906007,
      "grad_norm": 0.20813363790512085,
      "learning_rate": 0.00022040354826462666,
      "loss": 2.0191,
      "step": 1523
    },
    {
      "epoch": 0.6987620357634112,
      "grad_norm": 0.20448055863380432,
      "learning_rate": 0.0002197881362885426,
      "loss": 2.021,
      "step": 1524
    },
    {
      "epoch": 0.6992205410362219,
      "grad_norm": 0.20843841135501862,
      "learning_rate": 0.0002191733425643318,
      "loss": 1.9373,
      "step": 1525
    },
    {
      "epoch": 0.6996790463090325,
      "grad_norm": 0.20848727226257324,
      "learning_rate": 0.00021855916844845826,
      "loss": 1.9827,
      "step": 1526
    },
    {
      "epoch": 0.7001375515818432,
      "grad_norm": 0.19865581393241882,
      "learning_rate": 0.00021794561529601898,
      "loss": 1.9619,
      "step": 1527
    },
    {
      "epoch": 0.7005960568546539,
      "grad_norm": 0.21278177201747894,
      "learning_rate": 0.00021733268446074138,
      "loss": 1.9653,
      "step": 1528
    },
    {
      "epoch": 0.7010545621274644,
      "grad_norm": 0.2168363779783249,
      "learning_rate": 0.00021672037729497917,
      "loss": 1.9833,
      "step": 1529
    },
    {
      "epoch": 0.7015130674002751,
      "grad_norm": 0.2195511758327484,
      "learning_rate": 0.0002161086951497106,
      "loss": 2.0826,
      "step": 1530
    },
    {
      "epoch": 0.7019715726730857,
      "grad_norm": 0.2135552018880844,
      "learning_rate": 0.00021549763937453442,
      "loss": 2.1138,
      "step": 1531
    },
    {
      "epoch": 0.7024300779458964,
      "grad_norm": 0.20922909677028656,
      "learning_rate": 0.00021488721131766736,
      "loss": 2.0574,
      "step": 1532
    },
    {
      "epoch": 0.7028885832187071,
      "grad_norm": 0.20677387714385986,
      "learning_rate": 0.00021427741232594183,
      "loss": 1.9706,
      "step": 1533
    },
    {
      "epoch": 0.7033470884915176,
      "grad_norm": 0.20840921998023987,
      "learning_rate": 0.0002136682437448013,
      "loss": 2.0483,
      "step": 1534
    },
    {
      "epoch": 0.7038055937643283,
      "grad_norm": 0.21172036230564117,
      "learning_rate": 0.0002130597069182994,
      "loss": 1.9364,
      "step": 1535
    },
    {
      "epoch": 0.7042640990371389,
      "grad_norm": 0.21360346674919128,
      "learning_rate": 0.0002124518031890948,
      "loss": 2.0188,
      "step": 1536
    },
    {
      "epoch": 0.7047226043099496,
      "grad_norm": 0.21389159560203552,
      "learning_rate": 0.0002118445338984502,
      "loss": 2.0541,
      "step": 1537
    },
    {
      "epoch": 0.7051811095827603,
      "grad_norm": 0.2227029949426651,
      "learning_rate": 0.00021123790038622808,
      "loss": 2.0463,
      "step": 1538
    },
    {
      "epoch": 0.7056396148555708,
      "grad_norm": 0.21600441634655,
      "learning_rate": 0.0002106319039908879,
      "loss": 2.1504,
      "step": 1539
    },
    {
      "epoch": 0.7060981201283815,
      "grad_norm": 0.20475856959819794,
      "learning_rate": 0.00021002654604948412,
      "loss": 2.0354,
      "step": 1540
    },
    {
      "epoch": 0.7065566254011921,
      "grad_norm": 0.21849437057971954,
      "learning_rate": 0.00020942182789766172,
      "loss": 2.0231,
      "step": 1541
    },
    {
      "epoch": 0.7070151306740028,
      "grad_norm": 0.21484898030757904,
      "learning_rate": 0.00020881775086965492,
      "loss": 2.025,
      "step": 1542
    },
    {
      "epoch": 0.7074736359468133,
      "grad_norm": 0.22070999443531036,
      "learning_rate": 0.00020821431629828246,
      "loss": 2.0217,
      "step": 1543
    },
    {
      "epoch": 0.707932141219624,
      "grad_norm": 0.22360840439796448,
      "learning_rate": 0.00020761152551494643,
      "loss": 1.9796,
      "step": 1544
    },
    {
      "epoch": 0.7083906464924347,
      "grad_norm": 0.21885907649993896,
      "learning_rate": 0.00020700937984962798,
      "loss": 1.9894,
      "step": 1545
    },
    {
      "epoch": 0.7088491517652453,
      "grad_norm": 0.2167757749557495,
      "learning_rate": 0.0002064078806308848,
      "loss": 2.0886,
      "step": 1546
    },
    {
      "epoch": 0.709307657038056,
      "grad_norm": 0.2225911021232605,
      "learning_rate": 0.00020580702918584882,
      "loss": 2.1457,
      "step": 1547
    },
    {
      "epoch": 0.7097661623108665,
      "grad_norm": 0.20532001554965973,
      "learning_rate": 0.000205206826840222,
      "loss": 1.9099,
      "step": 1548
    },
    {
      "epoch": 0.7102246675836772,
      "grad_norm": 0.220353901386261,
      "learning_rate": 0.0002046072749182751,
      "loss": 2.1026,
      "step": 1549
    },
    {
      "epoch": 0.7106831728564879,
      "grad_norm": 0.20876623690128326,
      "learning_rate": 0.00020400837474284273,
      "loss": 2.0344,
      "step": 1550
    },
    {
      "epoch": 0.7111416781292985,
      "grad_norm": 0.22120672464370728,
      "learning_rate": 0.0002034101276353224,
      "loss": 2.0335,
      "step": 1551
    },
    {
      "epoch": 0.7116001834021092,
      "grad_norm": 0.2069578766822815,
      "learning_rate": 0.00020281253491567027,
      "loss": 2.0668,
      "step": 1552
    },
    {
      "epoch": 0.7120586886749197,
      "grad_norm": 0.20943233370780945,
      "learning_rate": 0.0002022155979023984,
      "loss": 1.9865,
      "step": 1553
    },
    {
      "epoch": 0.7125171939477304,
      "grad_norm": 0.21620026230812073,
      "learning_rate": 0.000201619317912573,
      "loss": 2.0349,
      "step": 1554
    },
    {
      "epoch": 0.712975699220541,
      "grad_norm": 0.21853972971439362,
      "learning_rate": 0.00020102369626180962,
      "loss": 1.9614,
      "step": 1555
    },
    {
      "epoch": 0.7134342044933517,
      "grad_norm": 0.20099127292633057,
      "learning_rate": 0.0002004287342642721,
      "loss": 1.9858,
      "step": 1556
    },
    {
      "epoch": 0.7138927097661623,
      "grad_norm": 0.2154405117034912,
      "learning_rate": 0.00019983443323266824,
      "loss": 2.1344,
      "step": 1557
    },
    {
      "epoch": 0.7143512150389729,
      "grad_norm": 0.20690707862377167,
      "learning_rate": 0.00019924079447824805,
      "loss": 2.0081,
      "step": 1558
    },
    {
      "epoch": 0.7148097203117836,
      "grad_norm": 0.2102072536945343,
      "learning_rate": 0.00019864781931079977,
      "loss": 1.984,
      "step": 1559
    },
    {
      "epoch": 0.7152682255845942,
      "grad_norm": 0.22246110439300537,
      "learning_rate": 0.00019805550903864773,
      "loss": 2.0107,
      "step": 1560
    },
    {
      "epoch": 0.7157267308574049,
      "grad_norm": 0.2070019245147705,
      "learning_rate": 0.00019746386496864948,
      "loss": 1.9997,
      "step": 1561
    },
    {
      "epoch": 0.7161852361302155,
      "grad_norm": 0.21238185465335846,
      "learning_rate": 0.00019687288840619226,
      "loss": 2.0138,
      "step": 1562
    },
    {
      "epoch": 0.7166437414030261,
      "grad_norm": 0.21874603629112244,
      "learning_rate": 0.0001962825806551911,
      "loss": 1.9751,
      "step": 1563
    },
    {
      "epoch": 0.7171022466758368,
      "grad_norm": 0.21486423909664154,
      "learning_rate": 0.0001956929430180846,
      "loss": 1.9703,
      "step": 1564
    },
    {
      "epoch": 0.7175607519486474,
      "grad_norm": 0.20629966259002686,
      "learning_rate": 0.00019510397679583374,
      "loss": 2.015,
      "step": 1565
    },
    {
      "epoch": 0.7180192572214581,
      "grad_norm": 0.21249209344387054,
      "learning_rate": 0.0001945156832879174,
      "loss": 1.9303,
      "step": 1566
    },
    {
      "epoch": 0.7184777624942686,
      "grad_norm": 0.20839910209178925,
      "learning_rate": 0.00019392806379233036,
      "loss": 2.0319,
      "step": 1567
    },
    {
      "epoch": 0.7189362677670793,
      "grad_norm": 0.21240414679050446,
      "learning_rate": 0.00019334111960558065,
      "loss": 1.9362,
      "step": 1568
    },
    {
      "epoch": 0.71939477303989,
      "grad_norm": 0.23237183690071106,
      "learning_rate": 0.00019275485202268573,
      "loss": 2.1019,
      "step": 1569
    },
    {
      "epoch": 0.7198532783127006,
      "grad_norm": 0.21136392652988434,
      "learning_rate": 0.00019216926233717085,
      "loss": 1.9137,
      "step": 1570
    },
    {
      "epoch": 0.7203117835855112,
      "grad_norm": 0.20676270127296448,
      "learning_rate": 0.00019158435184106498,
      "loss": 2.0483,
      "step": 1571
    },
    {
      "epoch": 0.7207702888583218,
      "grad_norm": 0.21879258751869202,
      "learning_rate": 0.00019100012182489905,
      "loss": 2.0003,
      "step": 1572
    },
    {
      "epoch": 0.7212287941311325,
      "grad_norm": 0.20756769180297852,
      "learning_rate": 0.00019041657357770226,
      "loss": 2.013,
      "step": 1573
    },
    {
      "epoch": 0.7216872994039432,
      "grad_norm": 0.22335326671600342,
      "learning_rate": 0.00018983370838699943,
      "loss": 2.0216,
      "step": 1574
    },
    {
      "epoch": 0.7221458046767538,
      "grad_norm": 0.2174278348684311,
      "learning_rate": 0.00018925152753880892,
      "loss": 1.9409,
      "step": 1575
    },
    {
      "epoch": 0.7226043099495644,
      "grad_norm": 0.20649616420269012,
      "learning_rate": 0.00018867003231763847,
      "loss": 2.0593,
      "step": 1576
    },
    {
      "epoch": 0.723062815222375,
      "grad_norm": 0.21547159552574158,
      "learning_rate": 0.00018808922400648375,
      "loss": 2.0081,
      "step": 1577
    },
    {
      "epoch": 0.7235213204951857,
      "grad_norm": 0.21018114686012268,
      "learning_rate": 0.00018750910388682428,
      "loss": 2.0163,
      "step": 1578
    },
    {
      "epoch": 0.7239798257679964,
      "grad_norm": 0.21254976093769073,
      "learning_rate": 0.00018692967323862125,
      "loss": 1.9657,
      "step": 1579
    },
    {
      "epoch": 0.724438331040807,
      "grad_norm": 0.218579962849617,
      "learning_rate": 0.00018635093334031517,
      "loss": 1.9849,
      "step": 1580
    },
    {
      "epoch": 0.7248968363136176,
      "grad_norm": 0.2116633951663971,
      "learning_rate": 0.00018577288546882165,
      "loss": 1.9349,
      "step": 1581
    },
    {
      "epoch": 0.7253553415864282,
      "grad_norm": 0.2190074920654297,
      "learning_rate": 0.00018519553089953023,
      "loss": 2.0223,
      "step": 1582
    },
    {
      "epoch": 0.7258138468592389,
      "grad_norm": 0.22413448989391327,
      "learning_rate": 0.0001846188709063001,
      "loss": 2.1032,
      "step": 1583
    },
    {
      "epoch": 0.7262723521320495,
      "grad_norm": 0.2222093641757965,
      "learning_rate": 0.00018404290676145857,
      "loss": 2.0567,
      "step": 1584
    },
    {
      "epoch": 0.7267308574048602,
      "grad_norm": 0.2129461169242859,
      "learning_rate": 0.00018346763973579722,
      "loss": 1.9659,
      "step": 1585
    },
    {
      "epoch": 0.7271893626776708,
      "grad_norm": 0.21499434113502502,
      "learning_rate": 0.00018289307109856939,
      "loss": 2.0036,
      "step": 1586
    },
    {
      "epoch": 0.7276478679504814,
      "grad_norm": 0.21895280480384827,
      "learning_rate": 0.0001823192021174882,
      "loss": 2.0389,
      "step": 1587
    },
    {
      "epoch": 0.7281063732232921,
      "grad_norm": 0.20958659052848816,
      "learning_rate": 0.0001817460340587223,
      "loss": 2.0692,
      "step": 1588
    },
    {
      "epoch": 0.7285648784961027,
      "grad_norm": 0.19955570995807648,
      "learning_rate": 0.00018117356818689445,
      "loss": 1.9274,
      "step": 1589
    },
    {
      "epoch": 0.7290233837689133,
      "grad_norm": 0.2103288173675537,
      "learning_rate": 0.00018060180576507756,
      "loss": 2.0281,
      "step": 1590
    },
    {
      "epoch": 0.729481889041724,
      "grad_norm": 0.21875479817390442,
      "learning_rate": 0.00018003074805479313,
      "loss": 2.0729,
      "step": 1591
    },
    {
      "epoch": 0.7299403943145346,
      "grad_norm": 0.2071479856967926,
      "learning_rate": 0.00017946039631600724,
      "loss": 1.979,
      "step": 1592
    },
    {
      "epoch": 0.7303988995873453,
      "grad_norm": 0.21238116919994354,
      "learning_rate": 0.00017889075180712837,
      "loss": 2.0806,
      "step": 1593
    },
    {
      "epoch": 0.7308574048601559,
      "grad_norm": 0.22257670760154724,
      "learning_rate": 0.00017832181578500512,
      "loss": 2.0105,
      "step": 1594
    },
    {
      "epoch": 0.7313159101329665,
      "grad_norm": 0.22076398134231567,
      "learning_rate": 0.0001777535895049221,
      "loss": 1.9527,
      "step": 1595
    },
    {
      "epoch": 0.7317744154057771,
      "grad_norm": 0.21873141825199127,
      "learning_rate": 0.0001771860742205988,
      "loss": 2.0589,
      "step": 1596
    },
    {
      "epoch": 0.7322329206785878,
      "grad_norm": 0.20710858702659607,
      "learning_rate": 0.00017661927118418525,
      "loss": 1.9674,
      "step": 1597
    },
    {
      "epoch": 0.7326914259513985,
      "grad_norm": 0.2243223637342453,
      "learning_rate": 0.00017605318164626066,
      "loss": 1.9954,
      "step": 1598
    },
    {
      "epoch": 0.7331499312242091,
      "grad_norm": 0.21919964253902435,
      "learning_rate": 0.00017548780685582949,
      "loss": 2.0927,
      "step": 1599
    },
    {
      "epoch": 0.7336084364970197,
      "grad_norm": 0.2128794938325882,
      "learning_rate": 0.00017492314806031922,
      "loss": 2.1009,
      "step": 1600
    },
    {
      "epoch": 0.7340669417698303,
      "grad_norm": 0.20615506172180176,
      "learning_rate": 0.00017435920650557806,
      "loss": 1.904,
      "step": 1601
    },
    {
      "epoch": 0.734525447042641,
      "grad_norm": 0.2088647484779358,
      "learning_rate": 0.00017379598343587112,
      "loss": 2.0111,
      "step": 1602
    },
    {
      "epoch": 0.7349839523154517,
      "grad_norm": 0.21245770156383514,
      "learning_rate": 0.00017323348009387878,
      "loss": 2.0964,
      "step": 1603
    },
    {
      "epoch": 0.7354424575882622,
      "grad_norm": 0.21076130867004395,
      "learning_rate": 0.0001726716977206929,
      "loss": 1.9937,
      "step": 1604
    },
    {
      "epoch": 0.7359009628610729,
      "grad_norm": 0.20355471968650818,
      "learning_rate": 0.00017211063755581525,
      "loss": 1.9673,
      "step": 1605
    },
    {
      "epoch": 0.7363594681338835,
      "grad_norm": 0.21622096002101898,
      "learning_rate": 0.0001715503008371536,
      "loss": 1.993,
      "step": 1606
    },
    {
      "epoch": 0.7368179734066942,
      "grad_norm": 0.21040453016757965,
      "learning_rate": 0.0001709906888010196,
      "loss": 2.0467,
      "step": 1607
    },
    {
      "epoch": 0.7372764786795049,
      "grad_norm": 0.21072076261043549,
      "learning_rate": 0.00017043180268212638,
      "loss": 2.0307,
      "step": 1608
    },
    {
      "epoch": 0.7377349839523154,
      "grad_norm": 0.2149541676044464,
      "learning_rate": 0.00016987364371358481,
      "loss": 2.0005,
      "step": 1609
    },
    {
      "epoch": 0.7381934892251261,
      "grad_norm": 0.22619763016700745,
      "learning_rate": 0.00016931621312690214,
      "loss": 2.0698,
      "step": 1610
    },
    {
      "epoch": 0.7386519944979367,
      "grad_norm": 0.2089175581932068,
      "learning_rate": 0.00016875951215197777,
      "loss": 2.0201,
      "step": 1611
    },
    {
      "epoch": 0.7391104997707474,
      "grad_norm": 0.22122539579868317,
      "learning_rate": 0.00016820354201710214,
      "loss": 2.0105,
      "step": 1612
    },
    {
      "epoch": 0.739569005043558,
      "grad_norm": 0.22035349905490875,
      "learning_rate": 0.00016764830394895203,
      "loss": 2.036,
      "step": 1613
    },
    {
      "epoch": 0.7400275103163686,
      "grad_norm": 0.22236762940883636,
      "learning_rate": 0.00016709379917259027,
      "loss": 2.0451,
      "step": 1614
    },
    {
      "epoch": 0.7404860155891793,
      "grad_norm": 0.20899082720279694,
      "learning_rate": 0.00016654002891146091,
      "loss": 1.9952,
      "step": 1615
    },
    {
      "epoch": 0.7409445208619899,
      "grad_norm": 0.21927949786186218,
      "learning_rate": 0.00016598699438738764,
      "loss": 1.9835,
      "step": 1616
    },
    {
      "epoch": 0.7414030261348006,
      "grad_norm": 0.2217399626970291,
      "learning_rate": 0.00016543469682057105,
      "loss": 2.0445,
      "step": 1617
    },
    {
      "epoch": 0.7418615314076112,
      "grad_norm": 0.21303631365299225,
      "learning_rate": 0.00016488313742958526,
      "loss": 2.0459,
      "step": 1618
    },
    {
      "epoch": 0.7423200366804218,
      "grad_norm": 0.21243950724601746,
      "learning_rate": 0.00016433231743137646,
      "loss": 2.0679,
      "step": 1619
    },
    {
      "epoch": 0.7427785419532325,
      "grad_norm": 0.2140006572008133,
      "learning_rate": 0.0001637822380412584,
      "loss": 1.9522,
      "step": 1620
    },
    {
      "epoch": 0.7432370472260431,
      "grad_norm": 0.2154926210641861,
      "learning_rate": 0.00016323290047291195,
      "loss": 1.9641,
      "step": 1621
    },
    {
      "epoch": 0.7436955524988538,
      "grad_norm": 0.22856086492538452,
      "learning_rate": 0.0001626843059383803,
      "loss": 2.0807,
      "step": 1622
    },
    {
      "epoch": 0.7441540577716643,
      "grad_norm": 0.22337085008621216,
      "learning_rate": 0.00016213645564806752,
      "loss": 2.0647,
      "step": 1623
    },
    {
      "epoch": 0.744612563044475,
      "grad_norm": 0.2250814139842987,
      "learning_rate": 0.0001615893508107359,
      "loss": 1.9152,
      "step": 1624
    },
    {
      "epoch": 0.7450710683172856,
      "grad_norm": 0.20749039947986603,
      "learning_rate": 0.00016104299263350252,
      "loss": 2.0484,
      "step": 1625
    },
    {
      "epoch": 0.7455295735900963,
      "grad_norm": 0.2096468210220337,
      "learning_rate": 0.00016049738232183758,
      "loss": 1.9655,
      "step": 1626
    },
    {
      "epoch": 0.745988078862907,
      "grad_norm": 0.2120881825685501,
      "learning_rate": 0.0001599525210795606,
      "loss": 1.918,
      "step": 1627
    },
    {
      "epoch": 0.7464465841357175,
      "grad_norm": 0.20597679913043976,
      "learning_rate": 0.00015940841010883889,
      "loss": 2.0039,
      "step": 1628
    },
    {
      "epoch": 0.7469050894085282,
      "grad_norm": 0.2197163999080658,
      "learning_rate": 0.00015886505061018413,
      "loss": 2.022,
      "step": 1629
    },
    {
      "epoch": 0.7473635946813388,
      "grad_norm": 0.21385709941387177,
      "learning_rate": 0.0001583224437824498,
      "loss": 1.9615,
      "step": 1630
    },
    {
      "epoch": 0.7478220999541495,
      "grad_norm": 0.20831948518753052,
      "learning_rate": 0.0001577805908228293,
      "loss": 2.0685,
      "step": 1631
    },
    {
      "epoch": 0.7482806052269602,
      "grad_norm": 0.22114643454551697,
      "learning_rate": 0.00015723949292685191,
      "loss": 1.9758,
      "step": 1632
    },
    {
      "epoch": 0.7487391104997707,
      "grad_norm": 0.22212569415569305,
      "learning_rate": 0.0001566991512883818,
      "loss": 1.9446,
      "step": 1633
    },
    {
      "epoch": 0.7491976157725814,
      "grad_norm": 0.22558395564556122,
      "learning_rate": 0.00015615956709961378,
      "loss": 2.1294,
      "step": 1634
    },
    {
      "epoch": 0.749656121045392,
      "grad_norm": 0.21286845207214355,
      "learning_rate": 0.00015562074155107215,
      "loss": 1.9869,
      "step": 1635
    },
    {
      "epoch": 0.7501146263182027,
      "grad_norm": 0.2169298380613327,
      "learning_rate": 0.0001550826758316068,
      "loss": 2.0541,
      "step": 1636
    },
    {
      "epoch": 0.7505731315910132,
      "grad_norm": 0.22596324980258942,
      "learning_rate": 0.00015454537112839122,
      "loss": 2.1018,
      "step": 1637
    },
    {
      "epoch": 0.7510316368638239,
      "grad_norm": 0.2097603976726532,
      "learning_rate": 0.00015400882862692033,
      "loss": 2.0481,
      "step": 1638
    },
    {
      "epoch": 0.7514901421366346,
      "grad_norm": 0.21506422758102417,
      "learning_rate": 0.00015347304951100665,
      "loss": 2.0244,
      "step": 1639
    },
    {
      "epoch": 0.7519486474094452,
      "grad_norm": 0.2103201001882553,
      "learning_rate": 0.00015293803496277907,
      "loss": 2.0718,
      "step": 1640
    },
    {
      "epoch": 0.7524071526822559,
      "grad_norm": 0.2119438648223877,
      "learning_rate": 0.00015240378616267886,
      "loss": 2.0304,
      "step": 1641
    },
    {
      "epoch": 0.7528656579550664,
      "grad_norm": 0.2066224366426468,
      "learning_rate": 0.00015187030428945843,
      "loss": 1.9136,
      "step": 1642
    },
    {
      "epoch": 0.7533241632278771,
      "grad_norm": 0.21281680464744568,
      "learning_rate": 0.0001513375905201776,
      "loss": 2.032,
      "step": 1643
    },
    {
      "epoch": 0.7537826685006878,
      "grad_norm": 0.20834359526634216,
      "learning_rate": 0.00015080564603020142,
      "loss": 2.0326,
      "step": 1644
    },
    {
      "epoch": 0.7542411737734984,
      "grad_norm": 0.22263237833976746,
      "learning_rate": 0.0001502744719931982,
      "loss": 1.9396,
      "step": 1645
    },
    {
      "epoch": 0.7546996790463091,
      "grad_norm": 0.20517906546592712,
      "learning_rate": 0.00014974406958113558,
      "loss": 2.0375,
      "step": 1646
    },
    {
      "epoch": 0.7551581843191196,
      "grad_norm": 0.2157820612192154,
      "learning_rate": 0.00014921443996427947,
      "loss": 1.9885,
      "step": 1647
    },
    {
      "epoch": 0.7556166895919303,
      "grad_norm": 0.218081533908844,
      "learning_rate": 0.0001486855843111901,
      "loss": 2.1159,
      "step": 1648
    },
    {
      "epoch": 0.756075194864741,
      "grad_norm": 0.21405591070652008,
      "learning_rate": 0.0001481575037887201,
      "loss": 2.0239,
      "step": 1649
    },
    {
      "epoch": 0.7565337001375516,
      "grad_norm": 0.21557943522930145,
      "learning_rate": 0.00014763019956201253,
      "loss": 2.0655,
      "step": 1650
    },
    {
      "epoch": 0.7569922054103623,
      "grad_norm": 0.21105466783046722,
      "learning_rate": 0.0001471036727944966,
      "loss": 1.9576,
      "step": 1651
    },
    {
      "epoch": 0.7574507106831728,
      "grad_norm": 0.21455194056034088,
      "learning_rate": 0.0001465779246478872,
      "loss": 1.9511,
      "step": 1652
    },
    {
      "epoch": 0.7579092159559835,
      "grad_norm": 0.20455604791641235,
      "learning_rate": 0.00014605295628218045,
      "loss": 1.9331,
      "step": 1653
    },
    {
      "epoch": 0.7583677212287941,
      "grad_norm": 0.21155160665512085,
      "learning_rate": 0.0001455287688556527,
      "loss": 1.9625,
      "step": 1654
    },
    {
      "epoch": 0.7588262265016048,
      "grad_norm": 0.2060501128435135,
      "learning_rate": 0.00014500536352485673,
      "loss": 1.9617,
      "step": 1655
    },
    {
      "epoch": 0.7592847317744154,
      "grad_norm": 0.2144545018672943,
      "learning_rate": 0.00014448274144461965,
      "loss": 1.9825,
      "step": 1656
    },
    {
      "epoch": 0.759743237047226,
      "grad_norm": 0.21097786724567413,
      "learning_rate": 0.00014396090376804112,
      "loss": 2.0484,
      "step": 1657
    },
    {
      "epoch": 0.7602017423200367,
      "grad_norm": 0.21261228621006012,
      "learning_rate": 0.00014343985164648926,
      "loss": 1.9252,
      "step": 1658
    },
    {
      "epoch": 0.7606602475928473,
      "grad_norm": 0.21784809231758118,
      "learning_rate": 0.00014291958622959973,
      "loss": 2.1581,
      "step": 1659
    },
    {
      "epoch": 0.761118752865658,
      "grad_norm": 0.21590444445610046,
      "learning_rate": 0.00014240010866527176,
      "loss": 2.0378,
      "step": 1660
    },
    {
      "epoch": 0.7615772581384685,
      "grad_norm": 0.20918042957782745,
      "learning_rate": 0.00014188142009966686,
      "loss": 2.0043,
      "step": 1661
    },
    {
      "epoch": 0.7620357634112792,
      "grad_norm": 0.22278830409049988,
      "learning_rate": 0.0001413635216772053,
      "loss": 1.9914,
      "step": 1662
    },
    {
      "epoch": 0.7624942686840899,
      "grad_norm": 0.21710264682769775,
      "learning_rate": 0.000140846414540564,
      "loss": 2.0377,
      "step": 1663
    },
    {
      "epoch": 0.7629527739569005,
      "grad_norm": 0.2116970419883728,
      "learning_rate": 0.00014033009983067452,
      "loss": 1.9911,
      "step": 1664
    },
    {
      "epoch": 0.7634112792297112,
      "grad_norm": 0.211927130818367,
      "learning_rate": 0.00013981457868671927,
      "loss": 1.969,
      "step": 1665
    },
    {
      "epoch": 0.7638697845025217,
      "grad_norm": 0.20697057247161865,
      "learning_rate": 0.0001392998522461305,
      "loss": 1.995,
      "step": 1666
    },
    {
      "epoch": 0.7643282897753324,
      "grad_norm": 0.2194519191980362,
      "learning_rate": 0.00013878592164458635,
      "loss": 2.0377,
      "step": 1667
    },
    {
      "epoch": 0.7647867950481431,
      "grad_norm": 0.22103075683116913,
      "learning_rate": 0.00013827278801600978,
      "loss": 1.981,
      "step": 1668
    },
    {
      "epoch": 0.7652453003209537,
      "grad_norm": 0.2139228731393814,
      "learning_rate": 0.0001377604524925647,
      "loss": 2.008,
      "step": 1669
    },
    {
      "epoch": 0.7657038055937643,
      "grad_norm": 0.21065013110637665,
      "learning_rate": 0.00013724891620465424,
      "loss": 1.9923,
      "step": 1670
    },
    {
      "epoch": 0.7661623108665749,
      "grad_norm": 0.21835178136825562,
      "learning_rate": 0.0001367381802809185,
      "loss": 2.0075,
      "step": 1671
    },
    {
      "epoch": 0.7666208161393856,
      "grad_norm": 0.23014241456985474,
      "learning_rate": 0.00013622824584823113,
      "loss": 2.0129,
      "step": 1672
    },
    {
      "epoch": 0.7670793214121963,
      "grad_norm": 0.21331645548343658,
      "learning_rate": 0.00013571911403169795,
      "loss": 2.0084,
      "step": 1673
    },
    {
      "epoch": 0.7675378266850069,
      "grad_norm": 0.2271205335855484,
      "learning_rate": 0.0001352107859546533,
      "loss": 2.0494,
      "step": 1674
    },
    {
      "epoch": 0.7679963319578175,
      "grad_norm": 0.213939368724823,
      "learning_rate": 0.00013470326273865886,
      "loss": 2.0328,
      "step": 1675
    },
    {
      "epoch": 0.7684548372306281,
      "grad_norm": 0.21541103720664978,
      "learning_rate": 0.00013419654550349985,
      "loss": 1.9425,
      "step": 1676
    },
    {
      "epoch": 0.7689133425034388,
      "grad_norm": 0.2088094800710678,
      "learning_rate": 0.00013369063536718346,
      "loss": 1.9766,
      "step": 1677
    },
    {
      "epoch": 0.7693718477762495,
      "grad_norm": 0.22045110166072845,
      "learning_rate": 0.00013318553344593632,
      "loss": 1.9719,
      "step": 1678
    },
    {
      "epoch": 0.7698303530490601,
      "grad_norm": 0.21845003962516785,
      "learning_rate": 0.00013268124085420136,
      "loss": 1.9786,
      "step": 1679
    },
    {
      "epoch": 0.7702888583218707,
      "grad_norm": 0.20567761361598969,
      "learning_rate": 0.0001321777587046364,
      "loss": 1.9899,
      "step": 1680
    },
    {
      "epoch": 0.7707473635946813,
      "grad_norm": 0.22374269366264343,
      "learning_rate": 0.00013167508810811059,
      "loss": 2.0154,
      "step": 1681
    },
    {
      "epoch": 0.771205868867492,
      "grad_norm": 0.2139168232679367,
      "learning_rate": 0.0001311732301737029,
      "loss": 2.0134,
      "step": 1682
    },
    {
      "epoch": 0.7716643741403026,
      "grad_norm": 0.20888787508010864,
      "learning_rate": 0.0001306721860086991,
      "loss": 2.0366,
      "step": 1683
    },
    {
      "epoch": 0.7721228794131133,
      "grad_norm": 0.208038330078125,
      "learning_rate": 0.00013017195671858928,
      "loss": 1.9687,
      "step": 1684
    },
    {
      "epoch": 0.7725813846859239,
      "grad_norm": 0.20752263069152832,
      "learning_rate": 0.0001296725434070661,
      "loss": 1.9598,
      "step": 1685
    },
    {
      "epoch": 0.7730398899587345,
      "grad_norm": 0.20393849909305573,
      "learning_rate": 0.00012917394717602121,
      "loss": 1.9986,
      "step": 1686
    },
    {
      "epoch": 0.7734983952315452,
      "grad_norm": 0.22817982733249664,
      "learning_rate": 0.00012867616912554426,
      "loss": 2.1076,
      "step": 1687
    },
    {
      "epoch": 0.7739569005043558,
      "grad_norm": 0.21016095578670502,
      "learning_rate": 0.00012817921035391882,
      "loss": 1.9958,
      "step": 1688
    },
    {
      "epoch": 0.7744154057771664,
      "grad_norm": 0.2149764448404312,
      "learning_rate": 0.00012768307195762168,
      "loss": 2.0193,
      "step": 1689
    },
    {
      "epoch": 0.774873911049977,
      "grad_norm": 0.22575724124908447,
      "learning_rate": 0.00012718775503131908,
      "loss": 1.9641,
      "step": 1690
    },
    {
      "epoch": 0.7753324163227877,
      "grad_norm": 0.21525436639785767,
      "learning_rate": 0.0001266932606678646,
      "loss": 1.9593,
      "step": 1691
    },
    {
      "epoch": 0.7757909215955984,
      "grad_norm": 0.21865308284759521,
      "learning_rate": 0.00012619958995829756,
      "loss": 1.985,
      "step": 1692
    },
    {
      "epoch": 0.776249426868409,
      "grad_norm": 0.21322782337665558,
      "learning_rate": 0.0001257067439918394,
      "loss": 1.9519,
      "step": 1693
    },
    {
      "epoch": 0.7767079321412196,
      "grad_norm": 0.21002227067947388,
      "learning_rate": 0.00012521472385589234,
      "loss": 2.0581,
      "step": 1694
    },
    {
      "epoch": 0.7771664374140302,
      "grad_norm": 0.2123413383960724,
      "learning_rate": 0.00012472353063603626,
      "loss": 1.9963,
      "step": 1695
    },
    {
      "epoch": 0.7776249426868409,
      "grad_norm": 0.22986717522144318,
      "learning_rate": 0.0001242331654160263,
      "loss": 2.0293,
      "step": 1696
    },
    {
      "epoch": 0.7780834479596516,
      "grad_norm": 0.2224844992160797,
      "learning_rate": 0.0001237436292777914,
      "loss": 2.0461,
      "step": 1697
    },
    {
      "epoch": 0.7785419532324622,
      "grad_norm": 0.20859204232692719,
      "learning_rate": 0.00012325492330143061,
      "loss": 2.0254,
      "step": 1698
    },
    {
      "epoch": 0.7790004585052728,
      "grad_norm": 0.21400931477546692,
      "learning_rate": 0.00012276704856521175,
      "loss": 2.0676,
      "step": 1699
    },
    {
      "epoch": 0.7794589637780834,
      "grad_norm": 0.21430806815624237,
      "learning_rate": 0.00012228000614556816,
      "loss": 1.9585,
      "step": 1700
    },
    {
      "epoch": 0.7799174690508941,
      "grad_norm": 0.2172161042690277,
      "learning_rate": 0.00012179379711709738,
      "loss": 1.9096,
      "step": 1701
    },
    {
      "epoch": 0.7803759743237048,
      "grad_norm": 0.21015897393226624,
      "learning_rate": 0.0001213084225525577,
      "loss": 2.1411,
      "step": 1702
    },
    {
      "epoch": 0.7808344795965153,
      "grad_norm": 0.20364156365394592,
      "learning_rate": 0.00012082388352286627,
      "loss": 2.0092,
      "step": 1703
    },
    {
      "epoch": 0.781292984869326,
      "grad_norm": 0.20627231895923615,
      "learning_rate": 0.00012034018109709716,
      "loss": 2.0081,
      "step": 1704
    },
    {
      "epoch": 0.7817514901421366,
      "grad_norm": 0.21120323240756989,
      "learning_rate": 0.00011985731634247809,
      "loss": 2.0784,
      "step": 1705
    },
    {
      "epoch": 0.7822099954149473,
      "grad_norm": 0.21254940330982208,
      "learning_rate": 0.00011937529032438904,
      "loss": 1.9975,
      "step": 1706
    },
    {
      "epoch": 0.782668500687758,
      "grad_norm": 0.2106795608997345,
      "learning_rate": 0.00011889410410635887,
      "loss": 1.9974,
      "step": 1707
    },
    {
      "epoch": 0.7831270059605685,
      "grad_norm": 0.20370933413505554,
      "learning_rate": 0.0001184137587500641,
      "loss": 2.0173,
      "step": 1708
    },
    {
      "epoch": 0.7835855112333792,
      "grad_norm": 0.22378553450107574,
      "learning_rate": 0.00011793425531532564,
      "loss": 1.9903,
      "step": 1709
    },
    {
      "epoch": 0.7840440165061898,
      "grad_norm": 0.2096746861934662,
      "learning_rate": 0.00011745559486010671,
      "loss": 2.113,
      "step": 1710
    },
    {
      "epoch": 0.7845025217790005,
      "grad_norm": 0.22115705907344818,
      "learning_rate": 0.00011697777844051105,
      "loss": 1.991,
      "step": 1711
    },
    {
      "epoch": 0.7849610270518111,
      "grad_norm": 0.20730508863925934,
      "learning_rate": 0.00011650080711077964,
      "loss": 2.0052,
      "step": 1712
    },
    {
      "epoch": 0.7854195323246217,
      "grad_norm": 0.21611107885837555,
      "learning_rate": 0.00011602468192328936,
      "loss": 2.0016,
      "step": 1713
    },
    {
      "epoch": 0.7858780375974324,
      "grad_norm": 0.21377933025360107,
      "learning_rate": 0.00011554940392854973,
      "loss": 2.013,
      "step": 1714
    },
    {
      "epoch": 0.786336542870243,
      "grad_norm": 0.21693699061870575,
      "learning_rate": 0.00011507497417520146,
      "loss": 2.05,
      "step": 1715
    },
    {
      "epoch": 0.7867950481430537,
      "grad_norm": 0.22443729639053345,
      "learning_rate": 0.00011460139371001339,
      "loss": 2.0254,
      "step": 1716
    },
    {
      "epoch": 0.7872535534158643,
      "grad_norm": 0.20095182955265045,
      "learning_rate": 0.00011412866357788049,
      "loss": 1.9987,
      "step": 1717
    },
    {
      "epoch": 0.7877120586886749,
      "grad_norm": 0.2166881412267685,
      "learning_rate": 0.00011365678482182207,
      "loss": 2.0521,
      "step": 1718
    },
    {
      "epoch": 0.7881705639614855,
      "grad_norm": 0.21766245365142822,
      "learning_rate": 0.0001131857584829783,
      "loss": 2.0133,
      "step": 1719
    },
    {
      "epoch": 0.7886290692342962,
      "grad_norm": 0.23814350366592407,
      "learning_rate": 0.0001127155856006093,
      "loss": 1.9859,
      "step": 1720
    },
    {
      "epoch": 0.7890875745071069,
      "grad_norm": 0.21680620312690735,
      "learning_rate": 0.00011224626721209141,
      "loss": 2.0366,
      "step": 1721
    },
    {
      "epoch": 0.7895460797799174,
      "grad_norm": 0.22185437381267548,
      "learning_rate": 0.0001117778043529164,
      "loss": 2.0451,
      "step": 1722
    },
    {
      "epoch": 0.7900045850527281,
      "grad_norm": 0.2212216854095459,
      "learning_rate": 0.0001113101980566879,
      "loss": 2.0259,
      "step": 1723
    },
    {
      "epoch": 0.7904630903255387,
      "grad_norm": 0.22035276889801025,
      "learning_rate": 0.00011084344935511958,
      "loss": 1.9714,
      "step": 1724
    },
    {
      "epoch": 0.7909215955983494,
      "grad_norm": 0.21128612756729126,
      "learning_rate": 0.00011037755927803345,
      "loss": 2.04,
      "step": 1725
    },
    {
      "epoch": 0.7913801008711601,
      "grad_norm": 0.21250511705875397,
      "learning_rate": 0.00010991252885335651,
      "loss": 2.0261,
      "step": 1726
    },
    {
      "epoch": 0.7918386061439706,
      "grad_norm": 0.21417243778705597,
      "learning_rate": 0.00010944835910711958,
      "loss": 1.9596,
      "step": 1727
    },
    {
      "epoch": 0.7922971114167813,
      "grad_norm": 0.21925154328346252,
      "learning_rate": 0.00010898505106345396,
      "loss": 2.0706,
      "step": 1728
    },
    {
      "epoch": 0.7927556166895919,
      "grad_norm": 0.21336814761161804,
      "learning_rate": 0.00010852260574459022,
      "loss": 1.9447,
      "step": 1729
    },
    {
      "epoch": 0.7932141219624026,
      "grad_norm": 0.2205604761838913,
      "learning_rate": 0.00010806102417085512,
      "loss": 2.04,
      "step": 1730
    },
    {
      "epoch": 0.7936726272352133,
      "grad_norm": 0.21539464592933655,
      "learning_rate": 0.00010760030736066951,
      "loss": 2.0493,
      "step": 1731
    },
    {
      "epoch": 0.7941311325080238,
      "grad_norm": 0.21450009942054749,
      "learning_rate": 0.00010714045633054687,
      "loss": 2.0348,
      "step": 1732
    },
    {
      "epoch": 0.7945896377808345,
      "grad_norm": 0.21369785070419312,
      "learning_rate": 0.00010668147209508971,
      "loss": 1.9559,
      "step": 1733
    },
    {
      "epoch": 0.7950481430536451,
      "grad_norm": 0.2226652055978775,
      "learning_rate": 0.00010622335566698877,
      "loss": 2.163,
      "step": 1734
    },
    {
      "epoch": 0.7955066483264558,
      "grad_norm": 0.20735086500644684,
      "learning_rate": 0.00010576610805701942,
      "loss": 2.0311,
      "step": 1735
    },
    {
      "epoch": 0.7959651535992663,
      "grad_norm": 0.22007210552692413,
      "learning_rate": 0.00010530973027404073,
      "loss": 1.9888,
      "step": 1736
    },
    {
      "epoch": 0.796423658872077,
      "grad_norm": 0.22485587000846863,
      "learning_rate": 0.00010485422332499212,
      "loss": 2.0204,
      "step": 1737
    },
    {
      "epoch": 0.7968821641448877,
      "grad_norm": 0.21233835816383362,
      "learning_rate": 0.00010439958821489165,
      "loss": 2.0404,
      "step": 1738
    },
    {
      "epoch": 0.7973406694176983,
      "grad_norm": 0.21458785235881805,
      "learning_rate": 0.00010394582594683428,
      "loss": 2.1134,
      "step": 1739
    },
    {
      "epoch": 0.797799174690509,
      "grad_norm": 0.21120969951152802,
      "learning_rate": 0.0001034929375219884,
      "loss": 1.9787,
      "step": 1740
    },
    {
      "epoch": 0.7982576799633195,
      "grad_norm": 0.22146326303482056,
      "learning_rate": 0.00010304092393959514,
      "loss": 1.9351,
      "step": 1741
    },
    {
      "epoch": 0.7987161852361302,
      "grad_norm": 0.2078889012336731,
      "learning_rate": 0.00010258978619696468,
      "loss": 2.0102,
      "step": 1742
    },
    {
      "epoch": 0.7991746905089409,
      "grad_norm": 0.2190144807100296,
      "learning_rate": 0.00010213952528947551,
      "loss": 2.0942,
      "step": 1743
    },
    {
      "epoch": 0.7996331957817515,
      "grad_norm": 0.21373145282268524,
      "learning_rate": 0.00010169014221057089,
      "loss": 1.9897,
      "step": 1744
    },
    {
      "epoch": 0.8000917010545622,
      "grad_norm": 0.21582815051078796,
      "learning_rate": 0.00010124163795175734,
      "loss": 1.99,
      "step": 1745
    },
    {
      "epoch": 0.8005502063273727,
      "grad_norm": 0.21146772801876068,
      "learning_rate": 0.00010079401350260287,
      "loss": 2.0547,
      "step": 1746
    },
    {
      "epoch": 0.8010087116001834,
      "grad_norm": 0.21636962890625,
      "learning_rate": 0.00010034726985073362,
      "loss": 2.0278,
      "step": 1747
    },
    {
      "epoch": 0.801467216872994,
      "grad_norm": 0.2111855000257492,
      "learning_rate": 9.9901407981833e-05,
      "loss": 2.0217,
      "step": 1748
    },
    {
      "epoch": 0.8019257221458047,
      "grad_norm": 0.20409125089645386,
      "learning_rate": 9.94564288796384e-05,
      "loss": 1.9398,
      "step": 1749
    },
    {
      "epoch": 0.8023842274186153,
      "grad_norm": 0.20008955895900726,
      "learning_rate": 9.901233352593953e-05,
      "loss": 1.9933,
      "step": 1750
    },
    {
      "epoch": 0.8028427326914259,
      "grad_norm": 0.21705128252506256,
      "learning_rate": 9.856912290057668e-05,
      "loss": 1.9997,
      "step": 1751
    },
    {
      "epoch": 0.8033012379642366,
      "grad_norm": 0.21685031056404114,
      "learning_rate": 9.812679798143748e-05,
      "loss": 2.0042,
      "step": 1752
    },
    {
      "epoch": 0.8037597432370472,
      "grad_norm": 0.22481180727481842,
      "learning_rate": 9.768535974445586e-05,
      "loss": 2.0428,
      "step": 1753
    },
    {
      "epoch": 0.8042182485098579,
      "grad_norm": 0.2321574091911316,
      "learning_rate": 9.724480916360906e-05,
      "loss": 2.0175,
      "step": 1754
    },
    {
      "epoch": 0.8046767537826685,
      "grad_norm": 0.22628888487815857,
      "learning_rate": 9.68051472109162e-05,
      "loss": 2.1118,
      "step": 1755
    },
    {
      "epoch": 0.8051352590554791,
      "grad_norm": 0.22457796335220337,
      "learning_rate": 9.636637485643529e-05,
      "loss": 1.9657,
      "step": 1756
    },
    {
      "epoch": 0.8055937643282898,
      "grad_norm": 0.21642456948757172,
      "learning_rate": 9.592849306826174e-05,
      "loss": 1.9967,
      "step": 1757
    },
    {
      "epoch": 0.8060522696011004,
      "grad_norm": 0.22476786375045776,
      "learning_rate": 9.549150281252633e-05,
      "loss": 2.0943,
      "step": 1758
    },
    {
      "epoch": 0.8065107748739111,
      "grad_norm": 0.24660331010818481,
      "learning_rate": 9.505540505339223e-05,
      "loss": 1.9462,
      "step": 1759
    },
    {
      "epoch": 0.8069692801467216,
      "grad_norm": 0.21481503546237946,
      "learning_rate": 9.4620200753054e-05,
      "loss": 2.0169,
      "step": 1760
    },
    {
      "epoch": 0.8074277854195323,
      "grad_norm": 0.21747137606143951,
      "learning_rate": 9.418589087173441e-05,
      "loss": 2.0231,
      "step": 1761
    },
    {
      "epoch": 0.807886290692343,
      "grad_norm": 0.21178947389125824,
      "learning_rate": 9.375247636768325e-05,
      "loss": 1.938,
      "step": 1762
    },
    {
      "epoch": 0.8083447959651536,
      "grad_norm": 0.22267508506774902,
      "learning_rate": 9.331995819717443e-05,
      "loss": 2.0417,
      "step": 1763
    },
    {
      "epoch": 0.8088033012379643,
      "grad_norm": 0.21884869039058685,
      "learning_rate": 9.288833731450419e-05,
      "loss": 1.9914,
      "step": 1764
    },
    {
      "epoch": 0.8092618065107748,
      "grad_norm": 0.21375304460525513,
      "learning_rate": 9.245761467198948e-05,
      "loss": 1.9935,
      "step": 1765
    },
    {
      "epoch": 0.8097203117835855,
      "grad_norm": 0.21833868324756622,
      "learning_rate": 9.20277912199648e-05,
      "loss": 1.9394,
      "step": 1766
    },
    {
      "epoch": 0.8101788170563962,
      "grad_norm": 0.22063380479812622,
      "learning_rate": 9.159886790678123e-05,
      "loss": 2.0157,
      "step": 1767
    },
    {
      "epoch": 0.8106373223292068,
      "grad_norm": 0.20850493013858795,
      "learning_rate": 9.11708456788033e-05,
      "loss": 1.9736,
      "step": 1768
    },
    {
      "epoch": 0.8110958276020174,
      "grad_norm": 0.21318040788173676,
      "learning_rate": 9.074372548040793e-05,
      "loss": 1.9372,
      "step": 1769
    },
    {
      "epoch": 0.811554332874828,
      "grad_norm": 0.21500568091869354,
      "learning_rate": 9.031750825398145e-05,
      "loss": 2.014,
      "step": 1770
    },
    {
      "epoch": 0.8120128381476387,
      "grad_norm": 0.21280577778816223,
      "learning_rate": 8.98921949399179e-05,
      "loss": 2.0452,
      "step": 1771
    },
    {
      "epoch": 0.8124713434204494,
      "grad_norm": 0.2111172378063202,
      "learning_rate": 8.94677864766173e-05,
      "loss": 1.9544,
      "step": 1772
    },
    {
      "epoch": 0.81292984869326,
      "grad_norm": 0.22338470816612244,
      "learning_rate": 8.904428380048269e-05,
      "loss": 1.9972,
      "step": 1773
    },
    {
      "epoch": 0.8133883539660706,
      "grad_norm": 0.22587040066719055,
      "learning_rate": 8.862168784591929e-05,
      "loss": 2.0558,
      "step": 1774
    },
    {
      "epoch": 0.8138468592388812,
      "grad_norm": 0.23038363456726074,
      "learning_rate": 8.819999954533115e-05,
      "loss": 1.9975,
      "step": 1775
    },
    {
      "epoch": 0.8143053645116919,
      "grad_norm": 0.20479805767536163,
      "learning_rate": 8.777921982911996e-05,
      "loss": 1.9177,
      "step": 1776
    },
    {
      "epoch": 0.8147638697845025,
      "grad_norm": 0.2203706055879593,
      "learning_rate": 8.735934962568253e-05,
      "loss": 1.9823,
      "step": 1777
    },
    {
      "epoch": 0.8152223750573132,
      "grad_norm": 0.21546988189220428,
      "learning_rate": 8.694038986140945e-05,
      "loss": 1.9887,
      "step": 1778
    },
    {
      "epoch": 0.8156808803301238,
      "grad_norm": 0.22397513687610626,
      "learning_rate": 8.652234146068206e-05,
      "loss": 2.0257,
      "step": 1779
    },
    {
      "epoch": 0.8161393856029344,
      "grad_norm": 0.21801428496837616,
      "learning_rate": 8.610520534587086e-05,
      "loss": 1.9422,
      "step": 1780
    },
    {
      "epoch": 0.8165978908757451,
      "grad_norm": 0.22166407108306885,
      "learning_rate": 8.568898243733397e-05,
      "loss": 1.9973,
      "step": 1781
    },
    {
      "epoch": 0.8170563961485557,
      "grad_norm": 0.22497108578681946,
      "learning_rate": 8.527367365341409e-05,
      "loss": 2.0483,
      "step": 1782
    },
    {
      "epoch": 0.8175149014213663,
      "grad_norm": 0.22142884135246277,
      "learning_rate": 8.485927991043757e-05,
      "loss": 2.106,
      "step": 1783
    },
    {
      "epoch": 0.817973406694177,
      "grad_norm": 0.22075378894805908,
      "learning_rate": 8.444580212271125e-05,
      "loss": 2.0364,
      "step": 1784
    },
    {
      "epoch": 0.8184319119669876,
      "grad_norm": 0.2093641608953476,
      "learning_rate": 8.403324120252159e-05,
      "loss": 2.048,
      "step": 1785
    },
    {
      "epoch": 0.8188904172397983,
      "grad_norm": 0.21878719329833984,
      "learning_rate": 8.362159806013175e-05,
      "loss": 1.9736,
      "step": 1786
    },
    {
      "epoch": 0.8193489225126089,
      "grad_norm": 0.22328490018844604,
      "learning_rate": 8.321087360377988e-05,
      "loss": 1.9863,
      "step": 1787
    },
    {
      "epoch": 0.8198074277854195,
      "grad_norm": 0.21988670527935028,
      "learning_rate": 8.280106873967752e-05,
      "loss": 2.0954,
      "step": 1788
    },
    {
      "epoch": 0.8202659330582301,
      "grad_norm": 0.2123972773551941,
      "learning_rate": 8.239218437200679e-05,
      "loss": 2.0112,
      "step": 1789
    },
    {
      "epoch": 0.8207244383310408,
      "grad_norm": 0.21902835369110107,
      "learning_rate": 8.198422140291939e-05,
      "loss": 1.9603,
      "step": 1790
    },
    {
      "epoch": 0.8211829436038515,
      "grad_norm": 0.21171347796916962,
      "learning_rate": 8.157718073253351e-05,
      "loss": 2.0621,
      "step": 1791
    },
    {
      "epoch": 0.8216414488766621,
      "grad_norm": 0.21930061280727386,
      "learning_rate": 8.117106325893287e-05,
      "loss": 1.9497,
      "step": 1792
    },
    {
      "epoch": 0.8220999541494727,
      "grad_norm": 0.21985183656215668,
      "learning_rate": 8.076586987816404e-05,
      "loss": 1.9823,
      "step": 1793
    },
    {
      "epoch": 0.8225584594222833,
      "grad_norm": 0.22060789167881012,
      "learning_rate": 8.036160148423449e-05,
      "loss": 2.0041,
      "step": 1794
    },
    {
      "epoch": 0.823016964695094,
      "grad_norm": 0.22023500502109528,
      "learning_rate": 7.995825896911141e-05,
      "loss": 2.0173,
      "step": 1795
    },
    {
      "epoch": 0.8234754699679047,
      "grad_norm": 0.20867501199245453,
      "learning_rate": 7.955584322271853e-05,
      "loss": 1.9949,
      "step": 1796
    },
    {
      "epoch": 0.8239339752407153,
      "grad_norm": 0.21236295998096466,
      "learning_rate": 7.915435513293523e-05,
      "loss": 2.0409,
      "step": 1797
    },
    {
      "epoch": 0.8243924805135259,
      "grad_norm": 0.2134450227022171,
      "learning_rate": 7.875379558559387e-05,
      "loss": 1.8902,
      "step": 1798
    },
    {
      "epoch": 0.8248509857863365,
      "grad_norm": 0.21306753158569336,
      "learning_rate": 7.835416546447838e-05,
      "loss": 1.9749,
      "step": 1799
    },
    {
      "epoch": 0.8253094910591472,
      "grad_norm": 0.2098451852798462,
      "learning_rate": 7.795546565132167e-05,
      "loss": 2.0274,
      "step": 1800
    },
    {
      "epoch": 0.8257679963319579,
      "grad_norm": 0.2244126796722412,
      "learning_rate": 7.755769702580412e-05,
      "loss": 2.0063,
      "step": 1801
    },
    {
      "epoch": 0.8262265016047684,
      "grad_norm": 0.21090929210186005,
      "learning_rate": 7.716086046555193e-05,
      "loss": 1.9981,
      "step": 1802
    },
    {
      "epoch": 0.8266850068775791,
      "grad_norm": 0.21630823612213135,
      "learning_rate": 7.676495684613432e-05,
      "loss": 2.0383,
      "step": 1803
    },
    {
      "epoch": 0.8271435121503897,
      "grad_norm": 0.21563266217708588,
      "learning_rate": 7.636998704106252e-05,
      "loss": 1.9367,
      "step": 1804
    },
    {
      "epoch": 0.8276020174232004,
      "grad_norm": 0.22116561233997345,
      "learning_rate": 7.597595192178702e-05,
      "loss": 1.9519,
      "step": 1805
    },
    {
      "epoch": 0.828060522696011,
      "grad_norm": 0.21977105736732483,
      "learning_rate": 7.558285235769646e-05,
      "loss": 1.9763,
      "step": 1806
    },
    {
      "epoch": 0.8285190279688216,
      "grad_norm": 0.21312573552131653,
      "learning_rate": 7.519068921611494e-05,
      "loss": 2.0335,
      "step": 1807
    },
    {
      "epoch": 0.8289775332416323,
      "grad_norm": 0.22274649143218994,
      "learning_rate": 7.479946336230047e-05,
      "loss": 2.0239,
      "step": 1808
    },
    {
      "epoch": 0.8294360385144429,
      "grad_norm": 0.22410453855991364,
      "learning_rate": 7.440917565944349e-05,
      "loss": 2.0393,
      "step": 1809
    },
    {
      "epoch": 0.8298945437872536,
      "grad_norm": 0.21330007910728455,
      "learning_rate": 7.4019826968664e-05,
      "loss": 1.9992,
      "step": 1810
    },
    {
      "epoch": 0.8303530490600642,
      "grad_norm": 0.21594150364398956,
      "learning_rate": 7.363141814901053e-05,
      "loss": 1.9891,
      "step": 1811
    },
    {
      "epoch": 0.8308115543328748,
      "grad_norm": 0.20046459138393402,
      "learning_rate": 7.32439500574577e-05,
      "loss": 2.0054,
      "step": 1812
    },
    {
      "epoch": 0.8312700596056855,
      "grad_norm": 0.2089758664369583,
      "learning_rate": 7.285742354890473e-05,
      "loss": 1.9273,
      "step": 1813
    },
    {
      "epoch": 0.8317285648784961,
      "grad_norm": 0.216389998793602,
      "learning_rate": 7.247183947617325e-05,
      "loss": 1.9532,
      "step": 1814
    },
    {
      "epoch": 0.8321870701513068,
      "grad_norm": 0.22176046669483185,
      "learning_rate": 7.20871986900053e-05,
      "loss": 1.9943,
      "step": 1815
    },
    {
      "epoch": 0.8326455754241173,
      "grad_norm": 0.21424785256385803,
      "learning_rate": 7.170350203906218e-05,
      "loss": 1.9687,
      "step": 1816
    },
    {
      "epoch": 0.833104080696928,
      "grad_norm": 0.2093542218208313,
      "learning_rate": 7.132075036992158e-05,
      "loss": 1.9995,
      "step": 1817
    },
    {
      "epoch": 0.8335625859697386,
      "grad_norm": 0.21715782582759857,
      "learning_rate": 7.093894452707666e-05,
      "loss": 1.9415,
      "step": 1818
    },
    {
      "epoch": 0.8340210912425493,
      "grad_norm": 0.21479164063930511,
      "learning_rate": 7.055808535293334e-05,
      "loss": 1.9587,
      "step": 1819
    },
    {
      "epoch": 0.83447959651536,
      "grad_norm": 0.22571024298667908,
      "learning_rate": 7.017817368780888e-05,
      "loss": 2.0178,
      "step": 1820
    },
    {
      "epoch": 0.8349381017881705,
      "grad_norm": 0.20602618157863617,
      "learning_rate": 6.979921036993042e-05,
      "loss": 1.9923,
      "step": 1821
    },
    {
      "epoch": 0.8353966070609812,
      "grad_norm": 0.2118396908044815,
      "learning_rate": 6.942119623543202e-05,
      "loss": 2.0642,
      "step": 1822
    },
    {
      "epoch": 0.8358551123337918,
      "grad_norm": 0.21564379334449768,
      "learning_rate": 6.904413211835414e-05,
      "loss": 2.0131,
      "step": 1823
    },
    {
      "epoch": 0.8363136176066025,
      "grad_norm": 0.2252080887556076,
      "learning_rate": 6.866801885064056e-05,
      "loss": 2.0389,
      "step": 1824
    },
    {
      "epoch": 0.8367721228794132,
      "grad_norm": 0.2115870714187622,
      "learning_rate": 6.829285726213769e-05,
      "loss": 1.9826,
      "step": 1825
    },
    {
      "epoch": 0.8372306281522237,
      "grad_norm": 0.23082710802555084,
      "learning_rate": 6.79186481805918e-05,
      "loss": 2.0532,
      "step": 1826
    },
    {
      "epoch": 0.8376891334250344,
      "grad_norm": 0.21160048246383667,
      "learning_rate": 6.754539243164754e-05,
      "loss": 1.9865,
      "step": 1827
    },
    {
      "epoch": 0.838147638697845,
      "grad_norm": 0.21186871826648712,
      "learning_rate": 6.717309083884654e-05,
      "loss": 2.0112,
      "step": 1828
    },
    {
      "epoch": 0.8386061439706557,
      "grad_norm": 0.21770504117012024,
      "learning_rate": 6.680174422362468e-05,
      "loss": 1.9508,
      "step": 1829
    },
    {
      "epoch": 0.8390646492434664,
      "grad_norm": 0.20482514798641205,
      "learning_rate": 6.643135340531136e-05,
      "loss": 2.0085,
      "step": 1830
    },
    {
      "epoch": 0.8395231545162769,
      "grad_norm": 0.21612153947353363,
      "learning_rate": 6.606191920112664e-05,
      "loss": 2.0941,
      "step": 1831
    },
    {
      "epoch": 0.8399816597890876,
      "grad_norm": 0.23194797337055206,
      "learning_rate": 6.569344242618036e-05,
      "loss": 1.9879,
      "step": 1832
    },
    {
      "epoch": 0.8404401650618982,
      "grad_norm": 0.23131988942623138,
      "learning_rate": 6.532592389346958e-05,
      "loss": 2.105,
      "step": 1833
    },
    {
      "epoch": 0.8408986703347089,
      "grad_norm": 0.20860381424427032,
      "learning_rate": 6.495936441387713e-05,
      "loss": 1.9435,
      "step": 1834
    },
    {
      "epoch": 0.8413571756075194,
      "grad_norm": 0.2117636650800705,
      "learning_rate": 6.459376479617013e-05,
      "loss": 1.9876,
      "step": 1835
    },
    {
      "epoch": 0.8418156808803301,
      "grad_norm": 0.21744032204151154,
      "learning_rate": 6.422912584699752e-05,
      "loss": 2.0385,
      "step": 1836
    },
    {
      "epoch": 0.8422741861531408,
      "grad_norm": 0.21241529285907745,
      "learning_rate": 6.386544837088904e-05,
      "loss": 1.965,
      "step": 1837
    },
    {
      "epoch": 0.8427326914259514,
      "grad_norm": 0.20964673161506653,
      "learning_rate": 6.350273317025251e-05,
      "loss": 1.9077,
      "step": 1838
    },
    {
      "epoch": 0.8431911966987621,
      "grad_norm": 0.21717427670955658,
      "learning_rate": 6.314098104537324e-05,
      "loss": 2.0036,
      "step": 1839
    },
    {
      "epoch": 0.8436497019715726,
      "grad_norm": 0.21879613399505615,
      "learning_rate": 6.278019279441122e-05,
      "loss": 2.0212,
      "step": 1840
    },
    {
      "epoch": 0.8441082072443833,
      "grad_norm": 0.22126102447509766,
      "learning_rate": 6.242036921339972e-05,
      "loss": 1.9895,
      "step": 1841
    },
    {
      "epoch": 0.844566712517194,
      "grad_norm": 0.21294614672660828,
      "learning_rate": 6.206151109624402e-05,
      "loss": 1.9903,
      "step": 1842
    },
    {
      "epoch": 0.8450252177900046,
      "grad_norm": 0.22445477545261383,
      "learning_rate": 6.170361923471868e-05,
      "loss": 1.9413,
      "step": 1843
    },
    {
      "epoch": 0.8454837230628153,
      "grad_norm": 0.21441799402236938,
      "learning_rate": 6.134669441846691e-05,
      "loss": 1.9694,
      "step": 1844
    },
    {
      "epoch": 0.8459422283356258,
      "grad_norm": 0.21734195947647095,
      "learning_rate": 6.099073743499772e-05,
      "loss": 2.0241,
      "step": 1845
    },
    {
      "epoch": 0.8464007336084365,
      "grad_norm": 0.22189994156360626,
      "learning_rate": 6.063574906968511e-05,
      "loss": 1.9672,
      "step": 1846
    },
    {
      "epoch": 0.8468592388812471,
      "grad_norm": 0.229840949177742,
      "learning_rate": 6.028173010576582e-05,
      "loss": 1.9338,
      "step": 1847
    },
    {
      "epoch": 0.8473177441540578,
      "grad_norm": 0.22288373112678528,
      "learning_rate": 5.9928681324337544e-05,
      "loss": 2.0254,
      "step": 1848
    },
    {
      "epoch": 0.8477762494268684,
      "grad_norm": 0.22014153003692627,
      "learning_rate": 5.957660350435773e-05,
      "loss": 1.9691,
      "step": 1849
    },
    {
      "epoch": 0.848234754699679,
      "grad_norm": 0.2213108390569687,
      "learning_rate": 5.922549742264122e-05,
      "loss": 2.0398,
      "step": 1850
    },
    {
      "epoch": 0.8486932599724897,
      "grad_norm": 0.21575769782066345,
      "learning_rate": 5.8875363853859166e-05,
      "loss": 1.9838,
      "step": 1851
    },
    {
      "epoch": 0.8491517652453003,
      "grad_norm": 0.20908822119235992,
      "learning_rate": 5.852620357053651e-05,
      "loss": 2.0225,
      "step": 1852
    },
    {
      "epoch": 0.849610270518111,
      "grad_norm": 0.22910979390144348,
      "learning_rate": 5.8178017343051336e-05,
      "loss": 1.9331,
      "step": 1853
    },
    {
      "epoch": 0.8500687757909215,
      "grad_norm": 0.21595340967178345,
      "learning_rate": 5.783080593963219e-05,
      "loss": 2.0015,
      "step": 1854
    },
    {
      "epoch": 0.8505272810637322,
      "grad_norm": 0.2159433811903,
      "learning_rate": 5.748457012635683e-05,
      "loss": 1.9956,
      "step": 1855
    },
    {
      "epoch": 0.8509857863365429,
      "grad_norm": 0.21303071081638336,
      "learning_rate": 5.713931066715078e-05,
      "loss": 1.902,
      "step": 1856
    },
    {
      "epoch": 0.8514442916093535,
      "grad_norm": 0.2087050825357437,
      "learning_rate": 5.679502832378497e-05,
      "loss": 2.0768,
      "step": 1857
    },
    {
      "epoch": 0.8519027968821642,
      "grad_norm": 0.2128382921218872,
      "learning_rate": 5.645172385587482e-05,
      "loss": 1.9397,
      "step": 1858
    },
    {
      "epoch": 0.8523613021549747,
      "grad_norm": 0.2117534875869751,
      "learning_rate": 5.6109398020877834e-05,
      "loss": 1.9262,
      "step": 1859
    },
    {
      "epoch": 0.8528198074277854,
      "grad_norm": 0.22160771489143372,
      "learning_rate": 5.576805157409265e-05,
      "loss": 1.9789,
      "step": 1860
    },
    {
      "epoch": 0.8532783127005961,
      "grad_norm": 0.2169644832611084,
      "learning_rate": 5.542768526865677e-05,
      "loss": 1.9429,
      "step": 1861
    },
    {
      "epoch": 0.8537368179734067,
      "grad_norm": 0.2240561842918396,
      "learning_rate": 5.508829985554509e-05,
      "loss": 2.0514,
      "step": 1862
    },
    {
      "epoch": 0.8541953232462174,
      "grad_norm": 0.21797043085098267,
      "learning_rate": 5.474989608356856e-05,
      "loss": 2.0458,
      "step": 1863
    },
    {
      "epoch": 0.8546538285190279,
      "grad_norm": 0.2090856283903122,
      "learning_rate": 5.441247469937194e-05,
      "loss": 1.9737,
      "step": 1864
    },
    {
      "epoch": 0.8551123337918386,
      "grad_norm": 0.21005772054195404,
      "learning_rate": 5.407603644743286e-05,
      "loss": 1.9623,
      "step": 1865
    },
    {
      "epoch": 0.8555708390646493,
      "grad_norm": 0.20576928555965424,
      "learning_rate": 5.374058207005944e-05,
      "loss": 1.8942,
      "step": 1866
    },
    {
      "epoch": 0.8560293443374599,
      "grad_norm": 0.22108455002307892,
      "learning_rate": 5.3406112307389066e-05,
      "loss": 1.9735,
      "step": 1867
    },
    {
      "epoch": 0.8564878496102705,
      "grad_norm": 0.218077152967453,
      "learning_rate": 5.3072627897386926e-05,
      "loss": 2.015,
      "step": 1868
    },
    {
      "epoch": 0.8569463548830811,
      "grad_norm": 0.21709051728248596,
      "learning_rate": 5.27401295758439e-05,
      "loss": 2.0023,
      "step": 1869
    },
    {
      "epoch": 0.8574048601558918,
      "grad_norm": 0.22305501997470856,
      "learning_rate": 5.2408618076375315e-05,
      "loss": 2.0758,
      "step": 1870
    },
    {
      "epoch": 0.8578633654287025,
      "grad_norm": 0.20986492931842804,
      "learning_rate": 5.207809413041914e-05,
      "loss": 2.0567,
      "step": 1871
    },
    {
      "epoch": 0.8583218707015131,
      "grad_norm": 0.20819778740406036,
      "learning_rate": 5.174855846723459e-05,
      "loss": 1.9639,
      "step": 1872
    },
    {
      "epoch": 0.8587803759743237,
      "grad_norm": 0.21768935024738312,
      "learning_rate": 5.1420011813900104e-05,
      "loss": 1.9938,
      "step": 1873
    },
    {
      "epoch": 0.8592388812471343,
      "grad_norm": 0.21815627813339233,
      "learning_rate": 5.109245489531211e-05,
      "loss": 1.9569,
      "step": 1874
    },
    {
      "epoch": 0.859697386519945,
      "grad_norm": 0.21125783026218414,
      "learning_rate": 5.0765888434183446e-05,
      "loss": 1.9118,
      "step": 1875
    },
    {
      "epoch": 0.8601558917927556,
      "grad_norm": 0.21321576833724976,
      "learning_rate": 5.0440313151041364e-05,
      "loss": 2.0146,
      "step": 1876
    },
    {
      "epoch": 0.8606143970655663,
      "grad_norm": 0.21765469014644623,
      "learning_rate": 5.011572976422657e-05,
      "loss": 1.9188,
      "step": 1877
    },
    {
      "epoch": 0.8610729023383769,
      "grad_norm": 0.21239829063415527,
      "learning_rate": 4.9792138989890825e-05,
      "loss": 1.9956,
      "step": 1878
    },
    {
      "epoch": 0.8615314076111875,
      "grad_norm": 0.22006644308567047,
      "learning_rate": 4.9469541541996234e-05,
      "loss": 1.9543,
      "step": 1879
    },
    {
      "epoch": 0.8619899128839982,
      "grad_norm": 0.2063804417848587,
      "learning_rate": 4.914793813231305e-05,
      "loss": 2.029,
      "step": 1880
    },
    {
      "epoch": 0.8624484181568088,
      "grad_norm": 0.2287290245294571,
      "learning_rate": 4.882732947041818e-05,
      "loss": 2.0255,
      "step": 1881
    },
    {
      "epoch": 0.8629069234296194,
      "grad_norm": 0.2139444202184677,
      "learning_rate": 4.850771626369416e-05,
      "loss": 2.0267,
      "step": 1882
    },
    {
      "epoch": 0.86336542870243,
      "grad_norm": 0.22281144559383392,
      "learning_rate": 4.818909921732662e-05,
      "loss": 1.9869,
      "step": 1883
    },
    {
      "epoch": 0.8638239339752407,
      "grad_norm": 0.2196056842803955,
      "learning_rate": 4.787147903430383e-05,
      "loss": 1.9542,
      "step": 1884
    },
    {
      "epoch": 0.8642824392480514,
      "grad_norm": 0.2277773916721344,
      "learning_rate": 4.755485641541424e-05,
      "loss": 1.985,
      "step": 1885
    },
    {
      "epoch": 0.864740944520862,
      "grad_norm": 0.21339185535907745,
      "learning_rate": 4.723923205924557e-05,
      "loss": 1.9376,
      "step": 1886
    },
    {
      "epoch": 0.8651994497936726,
      "grad_norm": 0.20995178818702698,
      "learning_rate": 4.6924606662182736e-05,
      "loss": 2.0219,
      "step": 1887
    },
    {
      "epoch": 0.8656579550664832,
      "grad_norm": 0.21453556418418884,
      "learning_rate": 4.6610980918406596e-05,
      "loss": 1.982,
      "step": 1888
    },
    {
      "epoch": 0.8661164603392939,
      "grad_norm": 0.22155457735061646,
      "learning_rate": 4.629835551989276e-05,
      "loss": 2.0296,
      "step": 1889
    },
    {
      "epoch": 0.8665749656121046,
      "grad_norm": 0.209043487906456,
      "learning_rate": 4.5986731156409224e-05,
      "loss": 1.9861,
      "step": 1890
    },
    {
      "epoch": 0.8670334708849152,
      "grad_norm": 0.21957042813301086,
      "learning_rate": 4.567610851551568e-05,
      "loss": 2.0845,
      "step": 1891
    },
    {
      "epoch": 0.8674919761577258,
      "grad_norm": 0.21351861953735352,
      "learning_rate": 4.536648828256146e-05,
      "loss": 1.9977,
      "step": 1892
    },
    {
      "epoch": 0.8679504814305364,
      "grad_norm": 0.21252942085266113,
      "learning_rate": 4.505787114068433e-05,
      "loss": 2.0087,
      "step": 1893
    },
    {
      "epoch": 0.8684089867033471,
      "grad_norm": 0.21305811405181885,
      "learning_rate": 4.4750257770808764e-05,
      "loss": 2.0396,
      "step": 1894
    },
    {
      "epoch": 0.8688674919761578,
      "grad_norm": 0.21838046610355377,
      "learning_rate": 4.444364885164448e-05,
      "loss": 1.9666,
      "step": 1895
    },
    {
      "epoch": 0.8693259972489684,
      "grad_norm": 0.2174653708934784,
      "learning_rate": 4.413804505968533e-05,
      "loss": 2.0025,
      "step": 1896
    },
    {
      "epoch": 0.869784502521779,
      "grad_norm": 0.21810795366764069,
      "learning_rate": 4.3833447069206944e-05,
      "loss": 2.072,
      "step": 1897
    },
    {
      "epoch": 0.8702430077945896,
      "grad_norm": 0.21322831511497498,
      "learning_rate": 4.352985555226635e-05,
      "loss": 2.0579,
      "step": 1898
    },
    {
      "epoch": 0.8707015130674003,
      "grad_norm": 0.21377459168434143,
      "learning_rate": 4.322727117869951e-05,
      "loss": 1.9616,
      "step": 1899
    },
    {
      "epoch": 0.871160018340211,
      "grad_norm": 0.2153778374195099,
      "learning_rate": 4.29256946161205e-05,
      "loss": 1.958,
      "step": 1900
    },
    {
      "epoch": 0.8716185236130215,
      "grad_norm": 0.20579054951667786,
      "learning_rate": 4.262512652991968e-05,
      "loss": 1.9291,
      "step": 1901
    },
    {
      "epoch": 0.8720770288858322,
      "grad_norm": 0.20723925530910492,
      "learning_rate": 4.2325567583262113e-05,
      "loss": 1.942,
      "step": 1902
    },
    {
      "epoch": 0.8725355341586428,
      "grad_norm": 0.21141906082630157,
      "learning_rate": 4.2027018437086895e-05,
      "loss": 1.9289,
      "step": 1903
    },
    {
      "epoch": 0.8729940394314535,
      "grad_norm": 0.2127448469400406,
      "learning_rate": 4.172947975010449e-05,
      "loss": 1.937,
      "step": 1904
    },
    {
      "epoch": 0.8734525447042641,
      "grad_norm": 0.22152157127857208,
      "learning_rate": 4.143295217879645e-05,
      "loss": 2.0742,
      "step": 1905
    },
    {
      "epoch": 0.8739110499770747,
      "grad_norm": 0.21782366931438446,
      "learning_rate": 4.113743637741296e-05,
      "loss": 2.0125,
      "step": 1906
    },
    {
      "epoch": 0.8743695552498854,
      "grad_norm": 0.20907482504844666,
      "learning_rate": 4.084293299797226e-05,
      "loss": 2.0136,
      "step": 1907
    },
    {
      "epoch": 0.874828060522696,
      "grad_norm": 0.22179867327213287,
      "learning_rate": 4.054944269025862e-05,
      "loss": 1.9327,
      "step": 1908
    },
    {
      "epoch": 0.8752865657955067,
      "grad_norm": 0.21399857103824615,
      "learning_rate": 4.025696610182095e-05,
      "loss": 1.935,
      "step": 1909
    },
    {
      "epoch": 0.8757450710683173,
      "grad_norm": 0.2262871265411377,
      "learning_rate": 3.996550387797187e-05,
      "loss": 1.9305,
      "step": 1910
    },
    {
      "epoch": 0.8762035763411279,
      "grad_norm": 0.2134685218334198,
      "learning_rate": 3.9675056661785556e-05,
      "loss": 1.9127,
      "step": 1911
    },
    {
      "epoch": 0.8766620816139385,
      "grad_norm": 0.20526432991027832,
      "learning_rate": 3.9385625094097154e-05,
      "loss": 2.0216,
      "step": 1912
    },
    {
      "epoch": 0.8771205868867492,
      "grad_norm": 0.20696309208869934,
      "learning_rate": 3.909720981350034e-05,
      "loss": 1.9222,
      "step": 1913
    },
    {
      "epoch": 0.8775790921595599,
      "grad_norm": 0.21824589371681213,
      "learning_rate": 3.880981145634704e-05,
      "loss": 2.0809,
      "step": 1914
    },
    {
      "epoch": 0.8780375974323704,
      "grad_norm": 0.2150755524635315,
      "learning_rate": 3.852343065674507e-05,
      "loss": 2.0104,
      "step": 1915
    },
    {
      "epoch": 0.8784961027051811,
      "grad_norm": 0.2185172587633133,
      "learning_rate": 3.8238068046557276e-05,
      "loss": 1.9481,
      "step": 1916
    },
    {
      "epoch": 0.8789546079779917,
      "grad_norm": 0.21892009675502777,
      "learning_rate": 3.795372425540006e-05,
      "loss": 1.9471,
      "step": 1917
    },
    {
      "epoch": 0.8794131132508024,
      "grad_norm": 0.2144346982240677,
      "learning_rate": 3.76703999106418e-05,
      "loss": 1.9389,
      "step": 1918
    },
    {
      "epoch": 0.8798716185236131,
      "grad_norm": 0.22050830721855164,
      "learning_rate": 3.7388095637401754e-05,
      "loss": 1.9242,
      "step": 1919
    },
    {
      "epoch": 0.8803301237964236,
      "grad_norm": 0.19997255504131317,
      "learning_rate": 3.7106812058548376e-05,
      "loss": 1.9405,
      "step": 1920
    },
    {
      "epoch": 0.8807886290692343,
      "grad_norm": 0.22276806831359863,
      "learning_rate": 3.682654979469807e-05,
      "loss": 1.9819,
      "step": 1921
    },
    {
      "epoch": 0.8812471343420449,
      "grad_norm": 0.21392330527305603,
      "learning_rate": 3.654730946421403e-05,
      "loss": 2.0348,
      "step": 1922
    },
    {
      "epoch": 0.8817056396148556,
      "grad_norm": 0.21851138770580292,
      "learning_rate": 3.6269091683204466e-05,
      "loss": 1.9954,
      "step": 1923
    },
    {
      "epoch": 0.8821641448876663,
      "grad_norm": 0.21530552208423615,
      "learning_rate": 3.5991897065521693e-05,
      "loss": 2.036,
      "step": 1924
    },
    {
      "epoch": 0.8826226501604768,
      "grad_norm": 0.21368546783924103,
      "learning_rate": 3.571572622276026e-05,
      "loss": 2.0255,
      "step": 1925
    },
    {
      "epoch": 0.8830811554332875,
      "grad_norm": 0.2265869379043579,
      "learning_rate": 3.544057976425619e-05,
      "loss": 2.1178,
      "step": 1926
    },
    {
      "epoch": 0.8835396607060981,
      "grad_norm": 0.21974653005599976,
      "learning_rate": 3.5166458297085146e-05,
      "loss": 2.0741,
      "step": 1927
    },
    {
      "epoch": 0.8839981659789088,
      "grad_norm": 0.21974951028823853,
      "learning_rate": 3.489336242606111e-05,
      "loss": 2.002,
      "step": 1928
    },
    {
      "epoch": 0.8844566712517193,
      "grad_norm": 0.20901674032211304,
      "learning_rate": 3.462129275373577e-05,
      "loss": 1.9811,
      "step": 1929
    },
    {
      "epoch": 0.88491517652453,
      "grad_norm": 0.20813800394535065,
      "learning_rate": 3.4350249880395924e-05,
      "loss": 2.0287,
      "step": 1930
    },
    {
      "epoch": 0.8853736817973407,
      "grad_norm": 0.2204243242740631,
      "learning_rate": 3.408023440406355e-05,
      "loss": 1.9668,
      "step": 1931
    },
    {
      "epoch": 0.8858321870701513,
      "grad_norm": 0.21631504595279694,
      "learning_rate": 3.381124692049331e-05,
      "loss": 2.0315,
      "step": 1932
    },
    {
      "epoch": 0.886290692342962,
      "grad_norm": 0.21561047434806824,
      "learning_rate": 3.354328802317197e-05,
      "loss": 1.982,
      "step": 1933
    },
    {
      "epoch": 0.8867491976157725,
      "grad_norm": 0.21339528262615204,
      "learning_rate": 3.327635830331677e-05,
      "loss": 1.9305,
      "step": 1934
    },
    {
      "epoch": 0.8872077028885832,
      "grad_norm": 0.21384020149707794,
      "learning_rate": 3.3010458349874206e-05,
      "loss": 1.999,
      "step": 1935
    },
    {
      "epoch": 0.8876662081613939,
      "grad_norm": 0.2145785093307495,
      "learning_rate": 3.2745588749518775e-05,
      "loss": 2.0385,
      "step": 1936
    },
    {
      "epoch": 0.8881247134342045,
      "grad_norm": 0.21353255212306976,
      "learning_rate": 3.248175008665161e-05,
      "loss": 1.9204,
      "step": 1937
    },
    {
      "epoch": 0.8885832187070152,
      "grad_norm": 0.21883352100849152,
      "learning_rate": 3.221894294339911e-05,
      "loss": 2.0217,
      "step": 1938
    },
    {
      "epoch": 0.8890417239798257,
      "grad_norm": 0.20873615145683289,
      "learning_rate": 3.1957167899611836e-05,
      "loss": 1.9525,
      "step": 1939
    },
    {
      "epoch": 0.8895002292526364,
      "grad_norm": 0.208133727312088,
      "learning_rate": 3.169642553286334e-05,
      "loss": 1.9929,
      "step": 1940
    },
    {
      "epoch": 0.889958734525447,
      "grad_norm": 0.214307501912117,
      "learning_rate": 3.143671641844831e-05,
      "loss": 2.0351,
      "step": 1941
    },
    {
      "epoch": 0.8904172397982577,
      "grad_norm": 0.2231110781431198,
      "learning_rate": 3.117804112938205e-05,
      "loss": 1.9461,
      "step": 1942
    },
    {
      "epoch": 0.8908757450710684,
      "grad_norm": 0.20967313647270203,
      "learning_rate": 3.092040023639869e-05,
      "loss": 2.069,
      "step": 1943
    },
    {
      "epoch": 0.8913342503438789,
      "grad_norm": 0.21966886520385742,
      "learning_rate": 3.066379430795002e-05,
      "loss": 2.0004,
      "step": 1944
    },
    {
      "epoch": 0.8917927556166896,
      "grad_norm": 0.21569140255451202,
      "learning_rate": 3.040822391020459e-05,
      "loss": 2.0273,
      "step": 1945
    },
    {
      "epoch": 0.8922512608895002,
      "grad_norm": 0.21958476305007935,
      "learning_rate": 3.0153689607045842e-05,
      "loss": 1.9481,
      "step": 1946
    },
    {
      "epoch": 0.8927097661623109,
      "grad_norm": 0.2143385261297226,
      "learning_rate": 2.9900191960071545e-05,
      "loss": 1.9734,
      "step": 1947
    },
    {
      "epoch": 0.8931682714351215,
      "grad_norm": 0.2196211814880371,
      "learning_rate": 2.9647731528591848e-05,
      "loss": 1.9779,
      "step": 1948
    },
    {
      "epoch": 0.8936267767079321,
      "grad_norm": 0.21265298128128052,
      "learning_rate": 2.9396308869628795e-05,
      "loss": 2.0082,
      "step": 1949
    },
    {
      "epoch": 0.8940852819807428,
      "grad_norm": 0.2225942462682724,
      "learning_rate": 2.914592453791448e-05,
      "loss": 2.0063,
      "step": 1950
    },
    {
      "epoch": 0.8945437872535534,
      "grad_norm": 0.21257665753364563,
      "learning_rate": 2.8896579085889994e-05,
      "loss": 2.0012,
      "step": 1951
    },
    {
      "epoch": 0.8950022925263641,
      "grad_norm": 0.2075442522764206,
      "learning_rate": 2.86482730637046e-05,
      "loss": 1.9642,
      "step": 1952
    },
    {
      "epoch": 0.8954607977991746,
      "grad_norm": 0.21527129411697388,
      "learning_rate": 2.840100701921383e-05,
      "loss": 2.0319,
      "step": 1953
    },
    {
      "epoch": 0.8959193030719853,
      "grad_norm": 0.21793372929096222,
      "learning_rate": 2.8154781497978898e-05,
      "loss": 2.035,
      "step": 1954
    },
    {
      "epoch": 0.896377808344796,
      "grad_norm": 0.21690189838409424,
      "learning_rate": 2.7909597043265013e-05,
      "loss": 1.9798,
      "step": 1955
    },
    {
      "epoch": 0.8968363136176066,
      "grad_norm": 0.2094467431306839,
      "learning_rate": 2.7665454196040662e-05,
      "loss": 1.9876,
      "step": 1956
    },
    {
      "epoch": 0.8972948188904173,
      "grad_norm": 0.21783050894737244,
      "learning_rate": 2.7422353494975905e-05,
      "loss": 2.0096,
      "step": 1957
    },
    {
      "epoch": 0.8977533241632278,
      "grad_norm": 0.21637599170207977,
      "learning_rate": 2.7180295476441573e-05,
      "loss": 1.9833,
      "step": 1958
    },
    {
      "epoch": 0.8982118294360385,
      "grad_norm": 0.22078004479408264,
      "learning_rate": 2.6939280674508016e-05,
      "loss": 1.9859,
      "step": 1959
    },
    {
      "epoch": 0.8986703347088492,
      "grad_norm": 0.2066592276096344,
      "learning_rate": 2.669930962094358e-05,
      "loss": 2.001,
      "step": 1960
    },
    {
      "epoch": 0.8991288399816598,
      "grad_norm": 0.2143029272556305,
      "learning_rate": 2.6460382845214126e-05,
      "loss": 2.073,
      "step": 1961
    },
    {
      "epoch": 0.8995873452544704,
      "grad_norm": 0.20978759229183197,
      "learning_rate": 2.6222500874481025e-05,
      "loss": 2.0483,
      "step": 1962
    },
    {
      "epoch": 0.900045850527281,
      "grad_norm": 0.22446830570697784,
      "learning_rate": 2.5985664233600827e-05,
      "loss": 2.0258,
      "step": 1963
    },
    {
      "epoch": 0.9005043558000917,
      "grad_norm": 0.2194334715604782,
      "learning_rate": 2.574987344512336e-05,
      "loss": 1.9603,
      "step": 1964
    },
    {
      "epoch": 0.9009628610729024,
      "grad_norm": 0.21245351433753967,
      "learning_rate": 2.5515129029290984e-05,
      "loss": 2.0254,
      "step": 1965
    },
    {
      "epoch": 0.901421366345713,
      "grad_norm": 0.22352994978427887,
      "learning_rate": 2.5281431504037556e-05,
      "loss": 1.987,
      "step": 1966
    },
    {
      "epoch": 0.9018798716185236,
      "grad_norm": 0.21166983246803284,
      "learning_rate": 2.504878138498684e-05,
      "loss": 1.9812,
      "step": 1967
    },
    {
      "epoch": 0.9023383768913342,
      "grad_norm": 0.21670743823051453,
      "learning_rate": 2.48171791854519e-05,
      "loss": 2.0365,
      "step": 1968
    },
    {
      "epoch": 0.9027968821641449,
      "grad_norm": 0.2184302657842636,
      "learning_rate": 2.4586625416433473e-05,
      "loss": 2.0205,
      "step": 1969
    },
    {
      "epoch": 0.9032553874369555,
      "grad_norm": 0.21107977628707886,
      "learning_rate": 2.435712058661921e-05,
      "loss": 2.0374,
      "step": 1970
    },
    {
      "epoch": 0.9037138927097662,
      "grad_norm": 0.221713587641716,
      "learning_rate": 2.4128665202382327e-05,
      "loss": 1.9679,
      "step": 1971
    },
    {
      "epoch": 0.9041723979825768,
      "grad_norm": 0.21069778501987457,
      "learning_rate": 2.3901259767780515e-05,
      "loss": 1.9704,
      "step": 1972
    },
    {
      "epoch": 0.9046309032553874,
      "grad_norm": 0.21514688432216644,
      "learning_rate": 2.367490478455514e-05,
      "loss": 2.0765,
      "step": 1973
    },
    {
      "epoch": 0.9050894085281981,
      "grad_norm": 0.21010364592075348,
      "learning_rate": 2.3449600752129597e-05,
      "loss": 1.9586,
      "step": 1974
    },
    {
      "epoch": 0.9055479138010087,
      "grad_norm": 0.21976174414157867,
      "learning_rate": 2.3225348167608685e-05,
      "loss": 1.9706,
      "step": 1975
    },
    {
      "epoch": 0.9060064190738194,
      "grad_norm": 0.21510343253612518,
      "learning_rate": 2.3002147525777118e-05,
      "loss": 2.044,
      "step": 1976
    },
    {
      "epoch": 0.90646492434663,
      "grad_norm": 0.21546922624111176,
      "learning_rate": 2.2779999319098856e-05,
      "loss": 1.9583,
      "step": 1977
    },
    {
      "epoch": 0.9069234296194406,
      "grad_norm": 0.22027602791786194,
      "learning_rate": 2.255890403771571e-05,
      "loss": 2.0145,
      "step": 1978
    },
    {
      "epoch": 0.9073819348922513,
      "grad_norm": 0.2219817042350769,
      "learning_rate": 2.233886216944614e-05,
      "loss": 1.9809,
      "step": 1979
    },
    {
      "epoch": 0.9078404401650619,
      "grad_norm": 0.21411274373531342,
      "learning_rate": 2.211987419978484e-05,
      "loss": 2.0601,
      "step": 1980
    },
    {
      "epoch": 0.9082989454378725,
      "grad_norm": 0.22263048589229584,
      "learning_rate": 2.1901940611900705e-05,
      "loss": 2.0603,
      "step": 1981
    },
    {
      "epoch": 0.9087574507106831,
      "grad_norm": 0.21497052907943726,
      "learning_rate": 2.168506188663666e-05,
      "loss": 2.0231,
      "step": 1982
    },
    {
      "epoch": 0.9092159559834938,
      "grad_norm": 0.21012760698795319,
      "learning_rate": 2.1469238502507925e-05,
      "loss": 2.0125,
      "step": 1983
    },
    {
      "epoch": 0.9096744612563045,
      "grad_norm": 0.2111775130033493,
      "learning_rate": 2.125447093570154e-05,
      "loss": 1.9683,
      "step": 1984
    },
    {
      "epoch": 0.9101329665291151,
      "grad_norm": 0.21622349321842194,
      "learning_rate": 2.1040759660074793e-05,
      "loss": 2.0317,
      "step": 1985
    },
    {
      "epoch": 0.9105914718019257,
      "grad_norm": 0.21961112320423126,
      "learning_rate": 2.0828105147154273e-05,
      "loss": 2.0,
      "step": 1986
    },
    {
      "epoch": 0.9110499770747363,
      "grad_norm": 0.2206241488456726,
      "learning_rate": 2.061650786613545e-05,
      "loss": 1.9797,
      "step": 1987
    },
    {
      "epoch": 0.911508482347547,
      "grad_norm": 0.22732362151145935,
      "learning_rate": 2.040596828388058e-05,
      "loss": 2.0273,
      "step": 1988
    },
    {
      "epoch": 0.9119669876203577,
      "grad_norm": 0.22434364259243011,
      "learning_rate": 2.019648686491865e-05,
      "loss": 1.9767,
      "step": 1989
    },
    {
      "epoch": 0.9124254928931683,
      "grad_norm": 0.22191472351551056,
      "learning_rate": 1.9988064071443767e-05,
      "loss": 1.9282,
      "step": 1990
    },
    {
      "epoch": 0.9128839981659789,
      "grad_norm": 0.22045555710792542,
      "learning_rate": 1.9780700363314253e-05,
      "loss": 2.0252,
      "step": 1991
    },
    {
      "epoch": 0.9133425034387895,
      "grad_norm": 0.20865285396575928,
      "learning_rate": 1.957439619805196e-05,
      "loss": 2.0015,
      "step": 1992
    },
    {
      "epoch": 0.9138010087116002,
      "grad_norm": 0.21824665367603302,
      "learning_rate": 1.9369152030840554e-05,
      "loss": 1.9346,
      "step": 1993
    },
    {
      "epoch": 0.9142595139844109,
      "grad_norm": 0.21625865995883942,
      "learning_rate": 1.916496831452552e-05,
      "loss": 2.0007,
      "step": 1994
    },
    {
      "epoch": 0.9147180192572214,
      "grad_norm": 0.21838118135929108,
      "learning_rate": 1.8961845499611998e-05,
      "loss": 1.8959,
      "step": 1995
    },
    {
      "epoch": 0.9151765245300321,
      "grad_norm": 0.2150646597146988,
      "learning_rate": 1.8759784034264925e-05,
      "loss": 2.026,
      "step": 1996
    },
    {
      "epoch": 0.9156350298028427,
      "grad_norm": 0.21855127811431885,
      "learning_rate": 1.855878436430708e-05,
      "loss": 2.0898,
      "step": 1997
    },
    {
      "epoch": 0.9160935350756534,
      "grad_norm": 0.21164308488368988,
      "learning_rate": 1.835884693321871e-05,
      "loss": 1.9616,
      "step": 1998
    },
    {
      "epoch": 0.916552040348464,
      "grad_norm": 0.22715087234973907,
      "learning_rate": 1.8159972182136386e-05,
      "loss": 2.1311,
      "step": 1999
    },
    {
      "epoch": 0.9170105456212746,
      "grad_norm": 0.22016048431396484,
      "learning_rate": 1.7962160549851945e-05,
      "loss": 2.0002,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 2181,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.426851506243502e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}