{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.2408963585434174,
  "eval_steps": 500,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0028011204481792717,
      "grad_norm": 0.296756774187088,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.1718,
      "step": 1
    },
    {
      "epoch": 0.0056022408963585435,
      "grad_norm": 0.36025628447532654,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.133,
      "step": 2
    },
    {
      "epoch": 0.008403361344537815,
      "grad_norm": 0.3665211796760559,
      "learning_rate": 2.5e-06,
      "loss": 0.1346,
      "step": 3
    },
    {
      "epoch": 0.011204481792717087,
      "grad_norm": 0.47092393040657043,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.2136,
      "step": 4
    },
    {
      "epoch": 0.014005602240896359,
      "grad_norm": 0.5220111608505249,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.2203,
      "step": 5
    },
    {
      "epoch": 0.01680672268907563,
      "grad_norm": 0.6161433458328247,
      "learning_rate": 5e-06,
      "loss": 0.0893,
      "step": 6
    },
    {
      "epoch": 0.0196078431372549,
      "grad_norm": 0.34641674160957336,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.1721,
      "step": 7
    },
    {
      "epoch": 0.022408963585434174,
      "grad_norm": 0.6245524287223816,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.0883,
      "step": 8
    },
    {
      "epoch": 0.025210084033613446,
      "grad_norm": 0.32467812299728394,
      "learning_rate": 7.5e-06,
      "loss": 0.167,
      "step": 9
    },
    {
      "epoch": 0.028011204481792718,
      "grad_norm": 0.7720906138420105,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.2531,
      "step": 10
    },
    {
      "epoch": 0.03081232492997199,
      "grad_norm": 0.38567203283309937,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.1309,
      "step": 11
    },
    {
      "epoch": 0.03361344537815126,
      "grad_norm": 0.5268588662147522,
      "learning_rate": 1e-05,
      "loss": 0.2119,
      "step": 12
    },
    {
      "epoch": 0.036414565826330535,
      "grad_norm": 0.3139660060405731,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 0.1741,
      "step": 13
    },
    {
      "epoch": 0.0392156862745098,
      "grad_norm": 0.35874423384666443,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.1337,
      "step": 14
    },
    {
      "epoch": 0.04201680672268908,
      "grad_norm": 0.3121141195297241,
      "learning_rate": 1.25e-05,
      "loss": 0.1745,
      "step": 15
    },
    {
      "epoch": 0.04481792717086835,
      "grad_norm": 0.4937639534473419,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.208,
      "step": 16
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 1.4663589000701904,
      "learning_rate": 1.4166666666666668e-05,
      "loss": 0.3469,
      "step": 17
    },
    {
      "epoch": 0.05042016806722689,
      "grad_norm": 0.3288300633430481,
      "learning_rate": 1.5e-05,
      "loss": 0.1735,
      "step": 18
    },
    {
      "epoch": 0.05322128851540616,
      "grad_norm": 0.7657830119132996,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.252,
      "step": 19
    },
    {
      "epoch": 0.056022408963585436,
      "grad_norm": 0.36719515919685364,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.1694,
      "step": 20
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 0.2963164448738098,
      "learning_rate": 1.75e-05,
      "loss": 0.1647,
      "step": 21
    },
    {
      "epoch": 0.06162464985994398,
      "grad_norm": 0.2958138883113861,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.1738,
      "step": 22
    },
    {
      "epoch": 0.06442577030812324,
      "grad_norm": 0.6379082202911377,
      "learning_rate": 1.9166666666666667e-05,
      "loss": 0.0961,
      "step": 23
    },
    {
      "epoch": 0.06722689075630252,
      "grad_norm": 0.40202125906944275,
      "learning_rate": 2e-05,
      "loss": 0.1353,
      "step": 24
    },
    {
      "epoch": 0.0700280112044818,
      "grad_norm": 0.631089448928833,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0944,
      "step": 25
    },
    {
      "epoch": 0.07282913165266107,
      "grad_norm": 0.6378069519996643,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 0.0931,
      "step": 26
    },
    {
      "epoch": 0.07563025210084033,
      "grad_norm": 0.3574202060699463,
      "learning_rate": 2.25e-05,
      "loss": 0.1729,
      "step": 27
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 0.3499799370765686,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.1732,
      "step": 28
    },
    {
      "epoch": 0.08123249299719888,
      "grad_norm": 0.5255885124206543,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.2186,
      "step": 29
    },
    {
      "epoch": 0.08403361344537816,
      "grad_norm": 0.35638874769210815,
      "learning_rate": 2.5e-05,
      "loss": 0.1338,
      "step": 30
    },
    {
      "epoch": 0.08683473389355742,
      "grad_norm": 0.3849603235721588,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 0.1326,
      "step": 31
    },
    {
      "epoch": 0.0896358543417367,
      "grad_norm": 0.32195600867271423,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.1715,
      "step": 32
    },
    {
      "epoch": 0.09243697478991597,
      "grad_norm": 0.8319142460823059,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.2664,
      "step": 33
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.7899373769760132,
      "learning_rate": 2.8333333333333335e-05,
      "loss": 0.2529,
      "step": 34
    },
    {
      "epoch": 0.09803921568627451,
      "grad_norm": 0.5739774107933044,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.0865,
      "step": 35
    },
    {
      "epoch": 0.10084033613445378,
      "grad_norm": 0.3802482783794403,
      "learning_rate": 3e-05,
      "loss": 0.131,
      "step": 36
    },
    {
      "epoch": 0.10364145658263306,
      "grad_norm": 0.4794873893260956,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 0.2136,
      "step": 37
    },
    {
      "epoch": 0.10644257703081232,
      "grad_norm": 0.3657197654247284,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.1318,
      "step": 38
    },
    {
      "epoch": 0.1092436974789916,
      "grad_norm": 0.3748391270637512,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.1775,
      "step": 39
    },
    {
      "epoch": 0.11204481792717087,
      "grad_norm": 0.4693293869495392,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.2114,
      "step": 40
    },
    {
      "epoch": 0.11484593837535013,
      "grad_norm": 0.3253132998943329,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 0.1775,
      "step": 41
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.5953426361083984,
      "learning_rate": 3.5e-05,
      "loss": 0.2213,
      "step": 42
    },
    {
      "epoch": 0.12044817927170869,
      "grad_norm": 0.3064591586589813,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 0.1328,
      "step": 43
    },
    {
      "epoch": 0.12324929971988796,
      "grad_norm": 0.37691906094551086,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.131,
      "step": 44
    },
    {
      "epoch": 0.12605042016806722,
      "grad_norm": 1.2265459299087524,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.3374,
      "step": 45
    },
    {
      "epoch": 0.12885154061624648,
      "grad_norm": 0.30182263255119324,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 0.1743,
      "step": 46
    },
    {
      "epoch": 0.13165266106442577,
      "grad_norm": 0.3886899948120117,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.131,
      "step": 47
    },
    {
      "epoch": 0.13445378151260504,
      "grad_norm": 0.5219215154647827,
      "learning_rate": 4e-05,
      "loss": 0.2145,
      "step": 48
    },
    {
      "epoch": 0.13725490196078433,
      "grad_norm": 0.2958211898803711,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 0.1715,
      "step": 49
    },
    {
      "epoch": 0.1400560224089636,
      "grad_norm": 0.290129154920578,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.1739,
      "step": 50
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.3508636951446533,
      "learning_rate": 4.25e-05,
      "loss": 0.132,
      "step": 51
    },
    {
      "epoch": 0.14565826330532214,
      "grad_norm": 0.6013919711112976,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 0.0944,
      "step": 52
    },
    {
      "epoch": 0.1484593837535014,
      "grad_norm": 0.5936094522476196,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 0.0949,
      "step": 53
    },
    {
      "epoch": 0.15126050420168066,
      "grad_norm": 0.38572150468826294,
      "learning_rate": 4.5e-05,
      "loss": 0.135,
      "step": 54
    },
    {
      "epoch": 0.15406162464985995,
      "grad_norm": 0.2600800693035126,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.1692,
      "step": 55
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 0.3701837360858917,
      "learning_rate": 4.666666666666667e-05,
      "loss": 0.1291,
      "step": 56
    },
    {
      "epoch": 0.15966386554621848,
      "grad_norm": 0.34277698397636414,
      "learning_rate": 4.75e-05,
      "loss": 0.1346,
      "step": 57
    },
    {
      "epoch": 0.16246498599439776,
      "grad_norm": 0.35784104466438293,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 0.177,
      "step": 58
    },
    {
      "epoch": 0.16526610644257703,
      "grad_norm": 0.8476487994194031,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 0.2581,
      "step": 59
    },
    {
      "epoch": 0.16806722689075632,
      "grad_norm": 0.8182944655418396,
      "learning_rate": 5e-05,
      "loss": 0.2639,
      "step": 60
    },
    {
      "epoch": 0.17086834733893558,
      "grad_norm": 0.29115599393844604,
      "learning_rate": 4.999995853979553e-05,
      "loss": 0.1691,
      "step": 61
    },
    {
      "epoch": 0.17366946778711484,
      "grad_norm": 0.7699853777885437,
      "learning_rate": 4.9999834159319655e-05,
      "loss": 0.264,
      "step": 62
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 0.7362309694290161,
      "learning_rate": 4.999962685898491e-05,
      "loss": 0.2508,
      "step": 63
    },
    {
      "epoch": 0.1792717086834734,
      "grad_norm": 0.4930090010166168,
      "learning_rate": 4.999933663947886e-05,
      "loss": 0.2159,
      "step": 64
    },
    {
      "epoch": 0.18207282913165265,
      "grad_norm": 0.3581504821777344,
      "learning_rate": 4.999896350176413e-05,
      "loss": 0.1278,
      "step": 65
    },
    {
      "epoch": 0.18487394957983194,
      "grad_norm": 0.5248085260391235,
      "learning_rate": 4.999850744707835e-05,
      "loss": 0.0941,
      "step": 66
    },
    {
      "epoch": 0.1876750700280112,
      "grad_norm": 0.3348913788795471,
      "learning_rate": 4.9997968476934156e-05,
      "loss": 0.1326,
      "step": 67
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.5374695658683777,
      "learning_rate": 4.999734659311921e-05,
      "loss": 0.095,
      "step": 68
    },
    {
      "epoch": 0.19327731092436976,
      "grad_norm": 0.28596481680870056,
      "learning_rate": 4.9996641797696206e-05,
      "loss": 0.1703,
      "step": 69
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 0.3090248703956604,
      "learning_rate": 4.999585409300281e-05,
      "loss": 0.1358,
      "step": 70
    },
    {
      "epoch": 0.19887955182072828,
      "grad_norm": 0.3411775827407837,
      "learning_rate": 4.999498348165169e-05,
      "loss": 0.1323,
      "step": 71
    },
    {
      "epoch": 0.20168067226890757,
      "grad_norm": 0.32754117250442505,
      "learning_rate": 4.999402996653051e-05,
      "loss": 0.1324,
      "step": 72
    },
    {
      "epoch": 0.20448179271708683,
      "grad_norm": 0.3174399137496948,
      "learning_rate": 4.9992993550801905e-05,
      "loss": 0.1304,
      "step": 73
    },
    {
      "epoch": 0.20728291316526612,
      "grad_norm": 0.286386102437973,
      "learning_rate": 4.999187423790347e-05,
      "loss": 0.1289,
      "step": 74
    },
    {
      "epoch": 0.21008403361344538,
      "grad_norm": 0.6507259607315063,
      "learning_rate": 4.999067203154777e-05,
      "loss": 0.2545,
      "step": 75
    },
    {
      "epoch": 0.21288515406162464,
      "grad_norm": 0.2852478623390198,
      "learning_rate": 4.998938693572229e-05,
      "loss": 0.1309,
      "step": 76
    },
    {
      "epoch": 0.21568627450980393,
      "grad_norm": 0.2769582271575928,
      "learning_rate": 4.9988018954689465e-05,
      "loss": 0.1694,
      "step": 77
    },
    {
      "epoch": 0.2184873949579832,
      "grad_norm": 0.3016033172607422,
      "learning_rate": 4.998656809298663e-05,
      "loss": 0.1752,
      "step": 78
    },
    {
      "epoch": 0.22128851540616246,
      "grad_norm": 0.2839262783527374,
      "learning_rate": 4.998503435542604e-05,
      "loss": 0.173,
      "step": 79
    },
    {
      "epoch": 0.22408963585434175,
      "grad_norm": 0.4902794659137726,
      "learning_rate": 4.9983417747094816e-05,
      "loss": 0.0811,
      "step": 80
    },
    {
      "epoch": 0.226890756302521,
      "grad_norm": 0.26926159858703613,
      "learning_rate": 4.998171827335494e-05,
      "loss": 0.1241,
      "step": 81
    },
    {
      "epoch": 0.22969187675070027,
      "grad_norm": 0.2879544198513031,
      "learning_rate": 4.997993593984327e-05,
      "loss": 0.1698,
      "step": 82
    },
    {
      "epoch": 0.23249299719887956,
      "grad_norm": 0.4781981110572815,
      "learning_rate": 4.997807075247146e-05,
      "loss": 0.0775,
      "step": 83
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.5029699206352234,
      "learning_rate": 4.997612271742601e-05,
      "loss": 0.215,
      "step": 84
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.27541235089302063,
      "learning_rate": 4.9974091841168195e-05,
      "loss": 0.1675,
      "step": 85
    },
    {
      "epoch": 0.24089635854341737,
      "grad_norm": 0.25892165303230286,
      "learning_rate": 4.997197813043404e-05,
      "loss": 0.1221,
      "step": 86
    },
    {
      "epoch": 0.24369747899159663,
      "grad_norm": 0.47153759002685547,
      "learning_rate": 4.996978159223436e-05,
      "loss": 0.0734,
      "step": 87
    },
    {
      "epoch": 0.24649859943977592,
      "grad_norm": 0.3439052402973175,
      "learning_rate": 4.9967502233854654e-05,
      "loss": 0.1698,
      "step": 88
    },
    {
      "epoch": 0.24929971988795518,
      "grad_norm": 1.2146350145339966,
      "learning_rate": 4.996514006285514e-05,
      "loss": 0.3275,
      "step": 89
    },
    {
      "epoch": 0.25210084033613445,
      "grad_norm": 0.5989933013916016,
      "learning_rate": 4.99626950870707e-05,
      "loss": 0.2192,
      "step": 90
    },
    {
      "epoch": 0.2549019607843137,
      "grad_norm": 0.481522798538208,
      "learning_rate": 4.996016731461088e-05,
      "loss": 0.0737,
      "step": 91
    },
    {
      "epoch": 0.25770308123249297,
      "grad_norm": 0.8134711384773254,
      "learning_rate": 4.995755675385982e-05,
      "loss": 0.2394,
      "step": 92
    },
    {
      "epoch": 0.2605042016806723,
      "grad_norm": 0.3162464201450348,
      "learning_rate": 4.995486341347629e-05,
      "loss": 0.1734,
      "step": 93
    },
    {
      "epoch": 0.26330532212885155,
      "grad_norm": 1.2472445964813232,
      "learning_rate": 4.99520873023936e-05,
      "loss": 0.331,
      "step": 94
    },
    {
      "epoch": 0.2661064425770308,
      "grad_norm": 0.4896290600299835,
      "learning_rate": 4.994922842981958e-05,
      "loss": 0.0775,
      "step": 95
    },
    {
      "epoch": 0.2689075630252101,
      "grad_norm": 0.6393519043922424,
      "learning_rate": 4.9946286805236616e-05,
      "loss": 0.2254,
      "step": 96
    },
    {
      "epoch": 0.27170868347338933,
      "grad_norm": 0.3101232945919037,
      "learning_rate": 4.994326243840153e-05,
      "loss": 0.1741,
      "step": 97
    },
    {
      "epoch": 0.27450980392156865,
      "grad_norm": 0.396138995885849,
      "learning_rate": 4.994015533934557e-05,
      "loss": 0.2115,
      "step": 98
    },
    {
      "epoch": 0.2773109243697479,
      "grad_norm": 0.4914068877696991,
      "learning_rate": 4.993696551837444e-05,
      "loss": 0.2207,
      "step": 99
    },
    {
      "epoch": 0.2801120448179272,
      "grad_norm": 0.31758832931518555,
      "learning_rate": 4.9933692986068165e-05,
      "loss": 0.1266,
      "step": 100
    },
    {
      "epoch": 0.28291316526610644,
      "grad_norm": 0.24751000106334686,
      "learning_rate": 4.993033775328115e-05,
      "loss": 0.1694,
      "step": 101
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.2383459210395813,
      "learning_rate": 4.992689983114209e-05,
      "loss": 0.17,
      "step": 102
    },
    {
      "epoch": 0.28851540616246496,
      "grad_norm": 0.3043301999568939,
      "learning_rate": 4.9923379231053925e-05,
      "loss": 0.1747,
      "step": 103
    },
    {
      "epoch": 0.2913165266106443,
      "grad_norm": 0.311767041683197,
      "learning_rate": 4.9919775964693854e-05,
      "loss": 0.1345,
      "step": 104
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 0.9271151423454285,
      "learning_rate": 4.991609004401324e-05,
      "loss": 0.3113,
      "step": 105
    },
    {
      "epoch": 0.2969187675070028,
      "grad_norm": 0.24133501946926117,
      "learning_rate": 4.991232148123761e-05,
      "loss": 0.1685,
      "step": 106
    },
    {
      "epoch": 0.29971988795518206,
      "grad_norm": 0.7193834781646729,
      "learning_rate": 4.99084702888666e-05,
      "loss": 0.2805,
      "step": 107
    },
    {
      "epoch": 0.3025210084033613,
      "grad_norm": 0.21335840225219727,
      "learning_rate": 4.990453647967389e-05,
      "loss": 0.1717,
      "step": 108
    },
    {
      "epoch": 0.30532212885154064,
      "grad_norm": 0.27446791529655457,
      "learning_rate": 4.9900520066707215e-05,
      "loss": 0.2081,
      "step": 109
    },
    {
      "epoch": 0.3081232492997199,
      "grad_norm": 0.5605149269104004,
      "learning_rate": 4.9896421063288286e-05,
      "loss": 0.1085,
      "step": 110
    },
    {
      "epoch": 0.31092436974789917,
      "grad_norm": 0.2565538287162781,
      "learning_rate": 4.989223948301273e-05,
      "loss": 0.1717,
      "step": 111
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 0.5685631632804871,
      "learning_rate": 4.988797533975009e-05,
      "loss": 0.1095,
      "step": 112
    },
    {
      "epoch": 0.3165266106442577,
      "grad_norm": 0.25811776518821716,
      "learning_rate": 4.9883628647643744e-05,
      "loss": 0.173,
      "step": 113
    },
    {
      "epoch": 0.31932773109243695,
      "grad_norm": 0.3600546717643738,
      "learning_rate": 4.9879199421110865e-05,
      "loss": 0.1432,
      "step": 114
    },
    {
      "epoch": 0.32212885154061627,
      "grad_norm": 0.22939907014369965,
      "learning_rate": 4.98746876748424e-05,
      "loss": 0.2009,
      "step": 115
    },
    {
      "epoch": 0.32492997198879553,
      "grad_norm": 0.3546235263347626,
      "learning_rate": 4.9870093423802964e-05,
      "loss": 0.1419,
      "step": 116
    },
    {
      "epoch": 0.3277310924369748,
      "grad_norm": 0.2727724015712738,
      "learning_rate": 4.986541668323086e-05,
      "loss": 0.2094,
      "step": 117
    },
    {
      "epoch": 0.33053221288515405,
      "grad_norm": 0.2383224368095398,
      "learning_rate": 4.986065746863797e-05,
      "loss": 0.1735,
      "step": 118
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.23489497601985931,
      "learning_rate": 4.985581579580973e-05,
      "loss": 0.1738,
      "step": 119
    },
    {
      "epoch": 0.33613445378151263,
      "grad_norm": 0.3256724178791046,
      "learning_rate": 4.985089168080509e-05,
      "loss": 0.1336,
      "step": 120
    },
    {
      "epoch": 0.3389355742296919,
      "grad_norm": 0.2290939837694168,
      "learning_rate": 4.9845885139956435e-05,
      "loss": 0.1732,
      "step": 121
    },
    {
      "epoch": 0.34173669467787116,
      "grad_norm": 0.2946733236312866,
      "learning_rate": 4.984079618986953e-05,
      "loss": 0.1377,
      "step": 122
    },
    {
      "epoch": 0.3445378151260504,
      "grad_norm": 0.9020264744758606,
      "learning_rate": 4.983562484742349e-05,
      "loss": 0.3229,
      "step": 123
    },
    {
      "epoch": 0.3473389355742297,
      "grad_norm": 0.2107078731060028,
      "learning_rate": 4.983037112977072e-05,
      "loss": 0.1738,
      "step": 124
    },
    {
      "epoch": 0.35014005602240894,
      "grad_norm": 0.21474942564964294,
      "learning_rate": 4.982503505433683e-05,
      "loss": 0.1698,
      "step": 125
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.3706744909286499,
      "learning_rate": 4.98196166388206e-05,
      "loss": 0.2061,
      "step": 126
    },
    {
      "epoch": 0.3557422969187675,
      "grad_norm": 0.19645719230175018,
      "learning_rate": 4.981411590119392e-05,
      "loss": 0.1687,
      "step": 127
    },
    {
      "epoch": 0.3585434173669468,
      "grad_norm": 0.4788144528865814,
      "learning_rate": 4.980853285970173e-05,
      "loss": 0.097,
      "step": 128
    },
    {
      "epoch": 0.36134453781512604,
      "grad_norm": 0.30618858337402344,
      "learning_rate": 4.980286753286195e-05,
      "loss": 0.1367,
      "step": 129
    },
    {
      "epoch": 0.3641456582633053,
      "grad_norm": 0.5136687159538269,
      "learning_rate": 4.979711993946543e-05,
      "loss": 0.2502,
      "step": 130
    },
    {
      "epoch": 0.36694677871148457,
      "grad_norm": 0.38218584656715393,
      "learning_rate": 4.9791290098575876e-05,
      "loss": 0.2166,
      "step": 131
    },
    {
      "epoch": 0.3697478991596639,
      "grad_norm": 0.3716464936733246,
      "learning_rate": 4.978537802952981e-05,
      "loss": 0.2071,
      "step": 132
    },
    {
      "epoch": 0.37254901960784315,
      "grad_norm": 0.21404078602790833,
      "learning_rate": 4.977938375193648e-05,
      "loss": 0.177,
      "step": 133
    },
    {
      "epoch": 0.3753501400560224,
      "grad_norm": 0.3054008483886719,
      "learning_rate": 4.9773307285677785e-05,
      "loss": 0.212,
      "step": 134
    },
    {
      "epoch": 0.37815126050420167,
      "grad_norm": 0.7401143312454224,
      "learning_rate": 4.976714865090827e-05,
      "loss": 0.278,
      "step": 135
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.3005881607532501,
      "learning_rate": 4.976090786805498e-05,
      "loss": 0.2057,
      "step": 136
    },
    {
      "epoch": 0.38375350140056025,
      "grad_norm": 1.065142035484314,
      "learning_rate": 4.975458495781746e-05,
      "loss": 0.3504,
      "step": 137
    },
    {
      "epoch": 0.3865546218487395,
      "grad_norm": 0.291591614484787,
      "learning_rate": 4.974817994116764e-05,
      "loss": 0.2105,
      "step": 138
    },
    {
      "epoch": 0.38935574229691877,
      "grad_norm": 0.21302799880504608,
      "learning_rate": 4.9741692839349765e-05,
      "loss": 0.1722,
      "step": 139
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.3730943500995636,
      "learning_rate": 4.973512367388038e-05,
      "loss": 0.2368,
      "step": 140
    },
    {
      "epoch": 0.3949579831932773,
      "grad_norm": 0.25385361909866333,
      "learning_rate": 4.9728472466548194e-05,
      "loss": 0.2002,
      "step": 141
    },
    {
      "epoch": 0.39775910364145656,
      "grad_norm": 0.25864890217781067,
      "learning_rate": 4.9721739239414034e-05,
      "loss": 0.1719,
      "step": 142
    },
    {
      "epoch": 0.4005602240896359,
      "grad_norm": 0.44104957580566406,
      "learning_rate": 4.971492401481079e-05,
      "loss": 0.1467,
      "step": 143
    },
    {
      "epoch": 0.40336134453781514,
      "grad_norm": 0.6197791695594788,
      "learning_rate": 4.9708026815343314e-05,
      "loss": 0.1173,
      "step": 144
    },
    {
      "epoch": 0.4061624649859944,
      "grad_norm": 0.6132116317749023,
      "learning_rate": 4.970104766388832e-05,
      "loss": 0.1187,
      "step": 145
    },
    {
      "epoch": 0.40896358543417366,
      "grad_norm": 0.28150931000709534,
      "learning_rate": 4.969398658359441e-05,
      "loss": 0.1739,
      "step": 146
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 0.41110119223594666,
      "learning_rate": 4.968684359788187e-05,
      "loss": 0.1441,
      "step": 147
    },
    {
      "epoch": 0.41456582633053224,
      "grad_norm": 0.2573298513889313,
      "learning_rate": 4.967961873044267e-05,
      "loss": 0.2043,
      "step": 148
    },
    {
      "epoch": 0.4173669467787115,
      "grad_norm": 0.5403711199760437,
      "learning_rate": 4.967231200524037e-05,
      "loss": 0.1091,
      "step": 149
    },
    {
      "epoch": 0.42016806722689076,
      "grad_norm": 0.25525861978530884,
      "learning_rate": 4.966492344651005e-05,
      "loss": 0.2077,
      "step": 150
    },
    {
      "epoch": 0.42296918767507,
      "grad_norm": 0.2501123547554016,
      "learning_rate": 4.965745307875819e-05,
      "loss": 0.2061,
      "step": 151
    },
    {
      "epoch": 0.4257703081232493,
      "grad_norm": 0.2235931009054184,
      "learning_rate": 4.964990092676263e-05,
      "loss": 0.1766,
      "step": 152
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.3031710386276245,
      "learning_rate": 4.9642267015572464e-05,
      "loss": 0.1371,
      "step": 153
    },
    {
      "epoch": 0.43137254901960786,
      "grad_norm": 0.48257577419281006,
      "learning_rate": 4.9634551370507986e-05,
      "loss": 0.0987,
      "step": 154
    },
    {
      "epoch": 0.4341736694677871,
      "grad_norm": 0.254314124584198,
      "learning_rate": 4.962675401716056e-05,
      "loss": 0.1627,
      "step": 155
    },
    {
      "epoch": 0.4369747899159664,
      "grad_norm": 0.46130886673927307,
      "learning_rate": 4.9618874981392596e-05,
      "loss": 0.0952,
      "step": 156
    },
    {
      "epoch": 0.43977591036414565,
      "grad_norm": 0.48739495873451233,
      "learning_rate": 4.961091428933738e-05,
      "loss": 0.2433,
      "step": 157
    },
    {
      "epoch": 0.4425770308123249,
      "grad_norm": 0.3604317307472229,
      "learning_rate": 4.96028719673991e-05,
      "loss": 0.2135,
      "step": 158
    },
    {
      "epoch": 0.44537815126050423,
      "grad_norm": 0.31823596358299255,
      "learning_rate": 4.959474804225263e-05,
      "loss": 0.209,
      "step": 159
    },
    {
      "epoch": 0.4481792717086835,
      "grad_norm": 0.4376252591609955,
      "learning_rate": 4.958654254084355e-05,
      "loss": 0.0907,
      "step": 160
    },
    {
      "epoch": 0.45098039215686275,
      "grad_norm": 0.5917844772338867,
      "learning_rate": 4.9578255490388007e-05,
      "loss": 0.2649,
      "step": 161
    },
    {
      "epoch": 0.453781512605042,
      "grad_norm": 0.33242353796958923,
      "learning_rate": 4.956988691837262e-05,
      "loss": 0.2071,
      "step": 162
    },
    {
      "epoch": 0.4565826330532213,
      "grad_norm": 0.19680732488632202,
      "learning_rate": 4.956143685255441e-05,
      "loss": 0.1706,
      "step": 163
    },
    {
      "epoch": 0.45938375350140054,
      "grad_norm": 0.19730661809444427,
      "learning_rate": 4.955290532096068e-05,
      "loss": 0.1656,
      "step": 164
    },
    {
      "epoch": 0.46218487394957986,
      "grad_norm": 0.24366319179534912,
      "learning_rate": 4.9544292351888966e-05,
      "loss": 0.129,
      "step": 165
    },
    {
      "epoch": 0.4649859943977591,
      "grad_norm": 0.20081248879432678,
      "learning_rate": 4.95355979739069e-05,
      "loss": 0.1659,
      "step": 166
    },
    {
      "epoch": 0.4677871148459384,
      "grad_norm": 0.24853447079658508,
      "learning_rate": 4.9526822215852145e-05,
      "loss": 0.1797,
      "step": 167
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.3966836929321289,
      "learning_rate": 4.951796510683227e-05,
      "loss": 0.2161,
      "step": 168
    },
    {
      "epoch": 0.4733893557422969,
      "grad_norm": 1.1678446531295776,
      "learning_rate": 4.950902667622468e-05,
      "loss": 0.3802,
      "step": 169
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.4717535674571991,
      "learning_rate": 4.95000069536765e-05,
      "loss": 0.222,
      "step": 170
    },
    {
      "epoch": 0.4789915966386555,
      "grad_norm": 0.2482030987739563,
      "learning_rate": 4.9490905969104514e-05,
      "loss": 0.1356,
      "step": 171
    },
    {
      "epoch": 0.48179271708683474,
      "grad_norm": 0.20962095260620117,
      "learning_rate": 4.9481723752695006e-05,
      "loss": 0.1644,
      "step": 172
    },
    {
      "epoch": 0.484593837535014,
      "grad_norm": 0.46536725759506226,
      "learning_rate": 4.9472460334903703e-05,
      "loss": 0.2479,
      "step": 173
    },
    {
      "epoch": 0.48739495798319327,
      "grad_norm": 0.4335561990737915,
      "learning_rate": 4.946311574645565e-05,
      "loss": 0.0947,
      "step": 174
    },
    {
      "epoch": 0.49019607843137253,
      "grad_norm": 0.4413890838623047,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 0.0965,
      "step": 175
    },
    {
      "epoch": 0.49299719887955185,
      "grad_norm": 0.17996694147586823,
      "learning_rate": 4.944418318183559e-05,
      "loss": 0.1686,
      "step": 176
    },
    {
      "epoch": 0.4957983193277311,
      "grad_norm": 0.17904537916183472,
      "learning_rate": 4.943459526845942e-05,
      "loss": 0.171,
      "step": 177
    },
    {
      "epoch": 0.49859943977591037,
      "grad_norm": 0.4312843382358551,
      "learning_rate": 4.9424926310017975e-05,
      "loss": 0.0961,
      "step": 178
    },
    {
      "epoch": 0.5014005602240896,
      "grad_norm": 0.27155429124832153,
      "learning_rate": 4.941517633858141e-05,
      "loss": 0.2079,
      "step": 179
    },
    {
      "epoch": 0.5042016806722689,
      "grad_norm": 0.2025201916694641,
      "learning_rate": 4.9405345386488614e-05,
      "loss": 0.1742,
      "step": 180
    },
    {
      "epoch": 0.5070028011204482,
      "grad_norm": 0.20862695574760437,
      "learning_rate": 4.939543348634703e-05,
      "loss": 0.1693,
      "step": 181
    },
    {
      "epoch": 0.5098039215686274,
      "grad_norm": 0.43479472398757935,
      "learning_rate": 4.9385440671032626e-05,
      "loss": 0.0942,
      "step": 182
    },
    {
      "epoch": 0.5126050420168067,
      "grad_norm": 0.2722318470478058,
      "learning_rate": 4.937536697368971e-05,
      "loss": 0.2083,
      "step": 183
    },
    {
      "epoch": 0.5154061624649859,
      "grad_norm": 0.4572315216064453,
      "learning_rate": 4.936521242773091e-05,
      "loss": 0.2492,
      "step": 184
    },
    {
      "epoch": 0.5182072829131653,
      "grad_norm": 0.5211023092269897,
      "learning_rate": 4.9354977066836986e-05,
      "loss": 0.2559,
      "step": 185
    },
    {
      "epoch": 0.5210084033613446,
      "grad_norm": 0.4220978319644928,
      "learning_rate": 4.934466092495673e-05,
      "loss": 0.0944,
      "step": 186
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 0.25524213910102844,
      "learning_rate": 4.9334264036306916e-05,
      "loss": 0.1332,
      "step": 187
    },
    {
      "epoch": 0.5266106442577031,
      "grad_norm": 0.1635034680366516,
      "learning_rate": 4.93237864353721e-05,
      "loss": 0.1691,
      "step": 188
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 0.1833125352859497,
      "learning_rate": 4.931322815690457e-05,
      "loss": 0.172,
      "step": 189
    },
    {
      "epoch": 0.5322128851540616,
      "grad_norm": 0.22643063962459564,
      "learning_rate": 4.930258923592418e-05,
      "loss": 0.1768,
      "step": 190
    },
    {
      "epoch": 0.5350140056022409,
      "grad_norm": 0.4311389923095703,
      "learning_rate": 4.9291869707718304e-05,
      "loss": 0.094,
      "step": 191
    },
    {
      "epoch": 0.5378151260504201,
      "grad_norm": 0.41472041606903076,
      "learning_rate": 4.9281069607841624e-05,
      "loss": 0.0918,
      "step": 192
    },
    {
      "epoch": 0.5406162464985994,
      "grad_norm": 0.19142650067806244,
      "learning_rate": 4.927018897211609e-05,
      "loss": 0.1737,
      "step": 193
    },
    {
      "epoch": 0.5434173669467787,
      "grad_norm": 0.27896803617477417,
      "learning_rate": 4.925922783663079e-05,
      "loss": 0.2096,
      "step": 194
    },
    {
      "epoch": 0.5462184873949579,
      "grad_norm": 0.2846761643886566,
      "learning_rate": 4.924818623774178e-05,
      "loss": 0.1254,
      "step": 195
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 0.1776442974805832,
      "learning_rate": 4.923706421207202e-05,
      "loss": 0.1715,
      "step": 196
    },
    {
      "epoch": 0.5518207282913166,
      "grad_norm": 0.2530427873134613,
      "learning_rate": 4.922586179651124e-05,
      "loss": 0.1286,
      "step": 197
    },
    {
      "epoch": 0.5546218487394958,
      "grad_norm": 0.3111589550971985,
      "learning_rate": 4.9214579028215776e-05,
      "loss": 0.209,
      "step": 198
    },
    {
      "epoch": 0.5574229691876751,
      "grad_norm": 0.4025686979293823,
      "learning_rate": 4.920321594460852e-05,
      "loss": 0.0844,
      "step": 199
    },
    {
      "epoch": 0.5602240896358543,
      "grad_norm": 0.4007948040962219,
      "learning_rate": 4.9191772583378705e-05,
      "loss": 0.0839,
      "step": 200
    },
    {
      "epoch": 0.5630252100840336,
      "grad_norm": 0.24552220106124878,
      "learning_rate": 4.9180248982481876e-05,
      "loss": 0.1228,
      "step": 201
    },
    {
      "epoch": 0.5658263305322129,
      "grad_norm": 0.3994346261024475,
      "learning_rate": 4.916864518013971e-05,
      "loss": 0.0792,
      "step": 202
    },
    {
      "epoch": 0.5686274509803921,
      "grad_norm": 0.24182085692882538,
      "learning_rate": 4.915696121483986e-05,
      "loss": 0.1216,
      "step": 203
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.6845493912696838,
      "learning_rate": 4.9145197125335916e-05,
      "loss": 0.271,
      "step": 204
    },
    {
      "epoch": 0.5742296918767507,
      "grad_norm": 0.35140976309776306,
      "learning_rate": 4.91333529506472e-05,
      "loss": 0.1824,
      "step": 205
    },
    {
      "epoch": 0.5770308123249299,
      "grad_norm": 0.2534499168395996,
      "learning_rate": 4.912142873005866e-05,
      "loss": 0.1623,
      "step": 206
    },
    {
      "epoch": 0.5798319327731093,
      "grad_norm": 0.21057064831256866,
      "learning_rate": 4.910942450312075e-05,
      "loss": 0.1185,
      "step": 207
    },
    {
      "epoch": 0.5826330532212886,
      "grad_norm": 0.2223557084798813,
      "learning_rate": 4.909734030964929e-05,
      "loss": 0.1259,
      "step": 208
    },
    {
      "epoch": 0.5854341736694678,
      "grad_norm": 0.2970251739025116,
      "learning_rate": 4.9085176189725314e-05,
      "loss": 0.1735,
      "step": 209
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.8583146929740906,
      "learning_rate": 4.907293218369499e-05,
      "loss": 0.2865,
      "step": 210
    },
    {
      "epoch": 0.5910364145658263,
      "grad_norm": 0.2618134617805481,
      "learning_rate": 4.906060833216942e-05,
      "loss": 0.1733,
      "step": 211
    },
    {
      "epoch": 0.5938375350140056,
      "grad_norm": 0.3876262605190277,
      "learning_rate": 4.904820467602458e-05,
      "loss": 0.0753,
      "step": 212
    },
    {
      "epoch": 0.5966386554621849,
      "grad_norm": 0.7885448932647705,
      "learning_rate": 4.90357212564011e-05,
      "loss": 0.2769,
      "step": 213
    },
    {
      "epoch": 0.5994397759103641,
      "grad_norm": 0.44926995038986206,
      "learning_rate": 4.9023158114704206e-05,
      "loss": 0.2188,
      "step": 214
    },
    {
      "epoch": 0.6022408963585434,
      "grad_norm": 1.0731271505355835,
      "learning_rate": 4.901051529260352e-05,
      "loss": 0.3633,
      "step": 215
    },
    {
      "epoch": 0.6050420168067226,
      "grad_norm": 0.38666459918022156,
      "learning_rate": 4.899779283203296e-05,
      "loss": 0.0786,
      "step": 216
    },
    {
      "epoch": 0.6078431372549019,
      "grad_norm": 0.40130844712257385,
      "learning_rate": 4.89849907751906e-05,
      "loss": 0.2174,
      "step": 217
    },
    {
      "epoch": 0.6106442577030813,
      "grad_norm": 0.22078561782836914,
      "learning_rate": 4.897210916453851e-05,
      "loss": 0.131,
      "step": 218
    },
    {
      "epoch": 0.6134453781512605,
      "grad_norm": 0.3916260600090027,
      "learning_rate": 4.895914804280262e-05,
      "loss": 0.0827,
      "step": 219
    },
    {
      "epoch": 0.6162464985994398,
      "grad_norm": 0.5899614095687866,
      "learning_rate": 4.89461074529726e-05,
      "loss": 0.2647,
      "step": 220
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 0.37688153982162476,
      "learning_rate": 4.893298743830168e-05,
      "loss": 0.2185,
      "step": 221
    },
    {
      "epoch": 0.6218487394957983,
      "grad_norm": 0.19744780659675598,
      "learning_rate": 4.891978804230656e-05,
      "loss": 0.1755,
      "step": 222
    },
    {
      "epoch": 0.6246498599439776,
      "grad_norm": 0.23267637193202972,
      "learning_rate": 4.890650930876719e-05,
      "loss": 0.1316,
      "step": 223
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 0.1993289738893509,
      "learning_rate": 4.889315128172669e-05,
      "loss": 0.1755,
      "step": 224
    },
    {
      "epoch": 0.6302521008403361,
      "grad_norm": 0.22977744042873383,
      "learning_rate": 4.88797140054912e-05,
      "loss": 0.1308,
      "step": 225
    },
    {
      "epoch": 0.6330532212885154,
      "grad_norm": 0.3395256996154785,
      "learning_rate": 4.88661975246297e-05,
      "loss": 0.2184,
      "step": 226
    },
    {
      "epoch": 0.6358543417366946,
      "grad_norm": 0.276090145111084,
      "learning_rate": 4.8852601883973846e-05,
      "loss": 0.1294,
      "step": 227
    },
    {
      "epoch": 0.6386554621848739,
      "grad_norm": 0.2874520421028137,
      "learning_rate": 4.883892712861791e-05,
      "loss": 0.2176,
      "step": 228
    },
    {
      "epoch": 0.6414565826330533,
      "grad_norm": 0.3347153663635254,
      "learning_rate": 4.882517330391854e-05,
      "loss": 0.2182,
      "step": 229
    },
    {
      "epoch": 0.6442577030812325,
      "grad_norm": 0.40832483768463135,
      "learning_rate": 4.8811340455494624e-05,
      "loss": 0.0955,
      "step": 230
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 0.2539953589439392,
      "learning_rate": 4.879742862922721e-05,
      "loss": 0.1308,
      "step": 231
    },
    {
      "epoch": 0.6498599439775911,
      "grad_norm": 0.26753100752830505,
      "learning_rate": 4.8783437871259254e-05,
      "loss": 0.2096,
      "step": 232
    },
    {
      "epoch": 0.6526610644257703,
      "grad_norm": 0.16798768937587738,
      "learning_rate": 4.876936822799553e-05,
      "loss": 0.1713,
      "step": 233
    },
    {
      "epoch": 0.6554621848739496,
      "grad_norm": 0.2817080020904541,
      "learning_rate": 4.875521974610247e-05,
      "loss": 0.2118,
      "step": 234
    },
    {
      "epoch": 0.6582633053221288,
      "grad_norm": 0.40791934728622437,
      "learning_rate": 4.874099247250798e-05,
      "loss": 0.0928,
      "step": 235
    },
    {
      "epoch": 0.6610644257703081,
      "grad_norm": 0.24818055331707,
      "learning_rate": 4.8726686454401325e-05,
      "loss": 0.2097,
      "step": 236
    },
    {
      "epoch": 0.6638655462184874,
      "grad_norm": 0.420173317193985,
      "learning_rate": 4.8712301739232935e-05,
      "loss": 0.2445,
      "step": 237
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.2433900535106659,
      "learning_rate": 4.869783837471427e-05,
      "loss": 0.1331,
      "step": 238
    },
    {
      "epoch": 0.6694677871148459,
      "grad_norm": 0.3094242811203003,
      "learning_rate": 4.8683296408817644e-05,
      "loss": 0.2125,
      "step": 239
    },
    {
      "epoch": 0.6722689075630253,
      "grad_norm": 0.24864792823791504,
      "learning_rate": 4.8668675889776095e-05,
      "loss": 0.1326,
      "step": 240
    },
    {
      "epoch": 0.6750700280112045,
      "grad_norm": 0.16234812140464783,
      "learning_rate": 4.8653976866083206e-05,
      "loss": 0.17,
      "step": 241
    },
    {
      "epoch": 0.6778711484593838,
      "grad_norm": 0.30067870020866394,
      "learning_rate": 4.863919938649293e-05,
      "loss": 0.206,
      "step": 242
    },
    {
      "epoch": 0.680672268907563,
      "grad_norm": 0.43154123425483704,
      "learning_rate": 4.862434350001945e-05,
      "loss": 0.2505,
      "step": 243
    },
    {
      "epoch": 0.6834733893557423,
      "grad_norm": 0.4572393596172333,
      "learning_rate": 4.860940925593703e-05,
      "loss": 0.2417,
      "step": 244
    },
    {
      "epoch": 0.6862745098039216,
      "grad_norm": 0.2604037821292877,
      "learning_rate": 4.85943967037798e-05,
      "loss": 0.1362,
      "step": 245
    },
    {
      "epoch": 0.6890756302521008,
      "grad_norm": 0.2577713429927826,
      "learning_rate": 4.857930589334164e-05,
      "loss": 0.1339,
      "step": 246
    },
    {
      "epoch": 0.6918767507002801,
      "grad_norm": 0.17431576550006866,
      "learning_rate": 4.8564136874676e-05,
      "loss": 0.1719,
      "step": 247
    },
    {
      "epoch": 0.6946778711484594,
      "grad_norm": 0.271586537361145,
      "learning_rate": 4.854888969809573e-05,
      "loss": 0.1349,
      "step": 248
    },
    {
      "epoch": 0.6974789915966386,
      "grad_norm": 0.2800721824169159,
      "learning_rate": 4.8533564414172915e-05,
      "loss": 0.1354,
      "step": 249
    },
    {
      "epoch": 0.7002801120448179,
      "grad_norm": 0.388351172208786,
      "learning_rate": 4.851816107373871e-05,
      "loss": 0.2419,
      "step": 250
    },
    {
      "epoch": 0.7030812324929971,
      "grad_norm": 0.2688753604888916,
      "learning_rate": 4.850267972788316e-05,
      "loss": 0.1368,
      "step": 251
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.43289467692375183,
      "learning_rate": 4.848712042795505e-05,
      "loss": 0.0977,
      "step": 252
    },
    {
      "epoch": 0.7086834733893558,
      "grad_norm": 0.3105364739894867,
      "learning_rate": 4.847148322556171e-05,
      "loss": 0.218,
      "step": 253
    },
    {
      "epoch": 0.711484593837535,
      "grad_norm": 0.44322580099105835,
      "learning_rate": 4.8455768172568886e-05,
      "loss": 0.2419,
      "step": 254
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.8140190839767456,
      "learning_rate": 4.843997532110051e-05,
      "loss": 0.3223,
      "step": 255
    },
    {
      "epoch": 0.7170868347338936,
      "grad_norm": 0.26502156257629395,
      "learning_rate": 4.842410472353858e-05,
      "loss": 0.1313,
      "step": 256
    },
    {
      "epoch": 0.7198879551820728,
      "grad_norm": 0.8722953200340271,
      "learning_rate": 4.840815643252294e-05,
      "loss": 0.3257,
      "step": 257
    },
    {
      "epoch": 0.7226890756302521,
      "grad_norm": 0.2759450376033783,
      "learning_rate": 4.839213050095116e-05,
      "loss": 0.1385,
      "step": 258
    },
    {
      "epoch": 0.7254901960784313,
      "grad_norm": 0.17727164924144745,
      "learning_rate": 4.83760269819783e-05,
      "loss": 0.1713,
      "step": 259
    },
    {
      "epoch": 0.7282913165266106,
      "grad_norm": 0.20785532891750336,
      "learning_rate": 4.835984592901678e-05,
      "loss": 0.1776,
      "step": 260
    },
    {
      "epoch": 0.7310924369747899,
      "grad_norm": 0.5836495757102966,
      "learning_rate": 4.834358739573618e-05,
      "loss": 0.2779,
      "step": 261
    },
    {
      "epoch": 0.7338935574229691,
      "grad_norm": 0.33494001626968384,
      "learning_rate": 4.8327251436063064e-05,
      "loss": 0.2319,
      "step": 262
    },
    {
      "epoch": 0.7366946778711485,
      "grad_norm": 0.29367175698280334,
      "learning_rate": 4.831083810418082e-05,
      "loss": 0.1383,
      "step": 263
    },
    {
      "epoch": 0.7394957983193278,
      "grad_norm": 0.31937041878700256,
      "learning_rate": 4.8294347454529445e-05,
      "loss": 0.1407,
      "step": 264
    },
    {
      "epoch": 0.742296918767507,
      "grad_norm": 0.48022958636283875,
      "learning_rate": 4.82777795418054e-05,
      "loss": 0.1067,
      "step": 265
    },
    {
      "epoch": 0.7450980392156863,
      "grad_norm": 0.48949047923088074,
      "learning_rate": 4.826113442096141e-05,
      "loss": 0.1079,
      "step": 266
    },
    {
      "epoch": 0.7478991596638656,
      "grad_norm": 0.5554994344711304,
      "learning_rate": 4.8244412147206284e-05,
      "loss": 0.2744,
      "step": 267
    },
    {
      "epoch": 0.7507002801120448,
      "grad_norm": 0.22763153910636902,
      "learning_rate": 4.822761277600474e-05,
      "loss": 0.2015,
      "step": 268
    },
    {
      "epoch": 0.7535014005602241,
      "grad_norm": 0.45823562145233154,
      "learning_rate": 4.821073636307719e-05,
      "loss": 0.1057,
      "step": 269
    },
    {
      "epoch": 0.7563025210084033,
      "grad_norm": 0.21038657426834106,
      "learning_rate": 4.819378296439961e-05,
      "loss": 0.1802,
      "step": 270
    },
    {
      "epoch": 0.7591036414565826,
      "grad_norm": 0.21373850107192993,
      "learning_rate": 4.8176752636203314e-05,
      "loss": 0.1706,
      "step": 271
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.2585920989513397,
      "learning_rate": 4.815964543497476e-05,
      "loss": 0.2091,
      "step": 272
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 0.4250545799732208,
      "learning_rate": 4.81424614174554e-05,
      "loss": 0.2504,
      "step": 273
    },
    {
      "epoch": 0.7675070028011205,
      "grad_norm": 0.350320041179657,
      "learning_rate": 4.8125200640641455e-05,
      "loss": 0.2373,
      "step": 274
    },
    {
      "epoch": 0.7703081232492998,
      "grad_norm": 0.2049403339624405,
      "learning_rate": 4.8107863161783773e-05,
      "loss": 0.1697,
      "step": 275
    },
    {
      "epoch": 0.773109243697479,
      "grad_norm": 0.20530298352241516,
      "learning_rate": 4.8090449038387564e-05,
      "loss": 0.1794,
      "step": 276
    },
    {
      "epoch": 0.7759103641456583,
      "grad_norm": 0.18878710269927979,
      "learning_rate": 4.8072958328212294e-05,
      "loss": 0.1672,
      "step": 277
    },
    {
      "epoch": 0.7787114845938375,
      "grad_norm": 0.23467376828193665,
      "learning_rate": 4.805539108927142e-05,
      "loss": 0.2107,
      "step": 278
    },
    {
      "epoch": 0.7815126050420168,
      "grad_norm": 0.3365757465362549,
      "learning_rate": 4.8037747379832265e-05,
      "loss": 0.1352,
      "step": 279
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 0.47438693046569824,
      "learning_rate": 4.8020027258415764e-05,
      "loss": 0.1032,
      "step": 280
    },
    {
      "epoch": 0.7871148459383753,
      "grad_norm": 0.29125773906707764,
      "learning_rate": 4.8002230783796315e-05,
      "loss": 0.1374,
      "step": 281
    },
    {
      "epoch": 0.7899159663865546,
      "grad_norm": 0.597076952457428,
      "learning_rate": 4.798435801500154e-05,
      "loss": 0.2789,
      "step": 282
    },
    {
      "epoch": 0.7927170868347339,
      "grad_norm": 0.40489983558654785,
      "learning_rate": 4.7966409011312145e-05,
      "loss": 0.2436,
      "step": 283
    },
    {
      "epoch": 0.7955182072829131,
      "grad_norm": 0.2765220105648041,
      "learning_rate": 4.7948383832261665e-05,
      "loss": 0.1397,
      "step": 284
    },
    {
      "epoch": 0.7983193277310925,
      "grad_norm": 0.28682276606559753,
      "learning_rate": 4.793028253763633e-05,
      "loss": 0.1353,
      "step": 285
    },
    {
      "epoch": 0.8011204481792717,
      "grad_norm": 0.2538602352142334,
      "learning_rate": 4.791210518747479e-05,
      "loss": 0.2152,
      "step": 286
    },
    {
      "epoch": 0.803921568627451,
      "grad_norm": 0.2573007643222809,
      "learning_rate": 4.7893851842067984e-05,
      "loss": 0.2034,
      "step": 287
    },
    {
      "epoch": 0.8067226890756303,
      "grad_norm": 0.2098461240530014,
      "learning_rate": 4.7875522561958906e-05,
      "loss": 0.1647,
      "step": 288
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 0.20104509592056274,
      "learning_rate": 4.785711740794242e-05,
      "loss": 0.1711,
      "step": 289
    },
    {
      "epoch": 0.8123249299719888,
      "grad_norm": 0.17355301976203918,
      "learning_rate": 4.783863644106502e-05,
      "loss": 0.1701,
      "step": 290
    },
    {
      "epoch": 0.8151260504201681,
      "grad_norm": 0.4620436728000641,
      "learning_rate": 4.782007972262471e-05,
      "loss": 0.1005,
      "step": 291
    },
    {
      "epoch": 0.8179271708683473,
      "grad_norm": 0.18848982453346252,
      "learning_rate": 4.7801447314170695e-05,
      "loss": 0.1764,
      "step": 292
    },
    {
      "epoch": 0.8207282913165266,
      "grad_norm": 0.2763855457305908,
      "learning_rate": 4.7782739277503266e-05,
      "loss": 0.1372,
      "step": 293
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.29234346747398376,
      "learning_rate": 4.776395567467353e-05,
      "loss": 0.1395,
      "step": 294
    },
    {
      "epoch": 0.8263305322128851,
      "grad_norm": 0.30519282817840576,
      "learning_rate": 4.7745096567983256e-05,
      "loss": 0.2119,
      "step": 295
    },
    {
      "epoch": 0.8291316526610645,
      "grad_norm": 0.16128459572792053,
      "learning_rate": 4.772616201998464e-05,
      "loss": 0.1722,
      "step": 296
    },
    {
      "epoch": 0.8319327731092437,
      "grad_norm": 0.43482664227485657,
      "learning_rate": 4.770715209348009e-05,
      "loss": 0.0938,
      "step": 297
    },
    {
      "epoch": 0.834733893557423,
      "grad_norm": 0.19980540871620178,
      "learning_rate": 4.768806685152206e-05,
      "loss": 0.1747,
      "step": 298
    },
    {
      "epoch": 0.8375350140056023,
      "grad_norm": 0.23939190804958344,
      "learning_rate": 4.766890635741278e-05,
      "loss": 0.1334,
      "step": 299
    },
    {
      "epoch": 0.8403361344537815,
      "grad_norm": 0.40733805298805237,
      "learning_rate": 4.76496706747041e-05,
      "loss": 0.0893,
      "step": 300
    },
    {
      "epoch": 0.8431372549019608,
      "grad_norm": 0.22405779361724854,
      "learning_rate": 4.763035986719722e-05,
      "loss": 0.1742,
      "step": 301
    },
    {
      "epoch": 0.84593837535014,
      "grad_norm": 0.23476524651050568,
      "learning_rate": 4.761097399894257e-05,
      "loss": 0.129,
      "step": 302
    },
    {
      "epoch": 0.8487394957983193,
      "grad_norm": 0.2294142097234726,
      "learning_rate": 4.7591513134239505e-05,
      "loss": 0.13,
      "step": 303
    },
    {
      "epoch": 0.8515406162464986,
      "grad_norm": 0.24955376982688904,
      "learning_rate": 4.757197733763615e-05,
      "loss": 0.1777,
      "step": 304
    },
    {
      "epoch": 0.8543417366946778,
      "grad_norm": 0.19835838675498962,
      "learning_rate": 4.7552366673929136e-05,
      "loss": 0.1683,
      "step": 305
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.2246866673231125,
      "learning_rate": 4.7532681208163444e-05,
      "loss": 0.167,
      "step": 306
    },
    {
      "epoch": 0.8599439775910365,
      "grad_norm": 0.2068600207567215,
      "learning_rate": 4.751292100563215e-05,
      "loss": 0.17,
      "step": 307
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 0.2158859223127365,
      "learning_rate": 4.7493086131876216e-05,
      "loss": 0.166,
      "step": 308
    },
    {
      "epoch": 0.865546218487395,
      "grad_norm": 0.44647109508514404,
      "learning_rate": 4.747317665268427e-05,
      "loss": 0.2246,
      "step": 309
    },
    {
      "epoch": 0.8683473389355743,
      "grad_norm": 0.38463255763053894,
      "learning_rate": 4.74531926340924e-05,
      "loss": 0.0782,
      "step": 310
    },
    {
      "epoch": 0.8711484593837535,
      "grad_norm": 0.20261278748512268,
      "learning_rate": 4.743313414238394e-05,
      "loss": 0.1246,
      "step": 311
    },
    {
      "epoch": 0.8739495798319328,
      "grad_norm": 0.24954895675182343,
      "learning_rate": 4.74130012440892e-05,
      "loss": 0.1735,
      "step": 312
    },
    {
      "epoch": 0.876750700280112,
      "grad_norm": 0.3798324167728424,
      "learning_rate": 4.7392794005985326e-05,
      "loss": 0.0757,
      "step": 313
    },
    {
      "epoch": 0.8795518207282913,
      "grad_norm": 0.1796898990869522,
      "learning_rate": 4.7372512495096e-05,
      "loss": 0.1273,
      "step": 314
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 0.675372838973999,
      "learning_rate": 4.735215677869128e-05,
      "loss": 0.2697,
      "step": 315
    },
    {
      "epoch": 0.8851540616246498,
      "grad_norm": 1.0348286628723145,
      "learning_rate": 4.733172692428734e-05,
      "loss": 0.3294,
      "step": 316
    },
    {
      "epoch": 0.8879551820728291,
      "grad_norm": 0.22474908828735352,
      "learning_rate": 4.731122299964625e-05,
      "loss": 0.1263,
      "step": 317
    },
    {
      "epoch": 0.8907563025210085,
      "grad_norm": 0.4402739107608795,
      "learning_rate": 4.7290645072775764e-05,
      "loss": 0.2248,
      "step": 318
    },
    {
      "epoch": 0.8935574229691877,
      "grad_norm": 0.22885151207447052,
      "learning_rate": 4.726999321192908e-05,
      "loss": 0.1709,
      "step": 319
    },
    {
      "epoch": 0.896358543417367,
      "grad_norm": 0.1985294669866562,
      "learning_rate": 4.7249267485604644e-05,
      "loss": 0.1238,
      "step": 320
    },
    {
      "epoch": 0.8991596638655462,
      "grad_norm": 0.3767232298851013,
      "learning_rate": 4.7228467962545866e-05,
      "loss": 0.0794,
      "step": 321
    },
    {
      "epoch": 0.9019607843137255,
      "grad_norm": 0.20609763264656067,
      "learning_rate": 4.720759471174096e-05,
      "loss": 0.1668,
      "step": 322
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 0.6048465371131897,
      "learning_rate": 4.7186647802422644e-05,
      "loss": 0.2622,
      "step": 323
    },
    {
      "epoch": 0.907563025210084,
      "grad_norm": 0.23643067479133606,
      "learning_rate": 4.7165627304068e-05,
      "loss": 0.1786,
      "step": 324
    },
    {
      "epoch": 0.9103641456582633,
      "grad_norm": 0.3819364011287689,
      "learning_rate": 4.714453328639814e-05,
      "loss": 0.0801,
      "step": 325
    },
    {
      "epoch": 0.9131652661064426,
      "grad_norm": 0.382179856300354,
      "learning_rate": 4.712336581937805e-05,
      "loss": 0.0812,
      "step": 326
    },
    {
      "epoch": 0.9159663865546218,
      "grad_norm": 0.20134811103343964,
      "learning_rate": 4.710212497321633e-05,
      "loss": 0.1285,
      "step": 327
    },
    {
      "epoch": 0.9187675070028011,
      "grad_norm": 0.3812258541584015,
      "learning_rate": 4.7080810818364974e-05,
      "loss": 0.0812,
      "step": 328
    },
    {
      "epoch": 0.9215686274509803,
      "grad_norm": 0.6299641132354736,
      "learning_rate": 4.7059423425519105e-05,
      "loss": 0.2613,
      "step": 329
    },
    {
      "epoch": 0.9243697478991597,
      "grad_norm": 0.20863133668899536,
      "learning_rate": 4.703796286561679e-05,
      "loss": 0.1234,
      "step": 330
    },
    {
      "epoch": 0.927170868347339,
      "grad_norm": 0.5638542175292969,
      "learning_rate": 4.7016429209838764e-05,
      "loss": 0.2572,
      "step": 331
    },
    {
      "epoch": 0.9299719887955182,
      "grad_norm": 0.8459420800209045,
      "learning_rate": 4.6994822529608204e-05,
      "loss": 0.31,
      "step": 332
    },
    {
      "epoch": 0.9327731092436975,
      "grad_norm": 0.20429813861846924,
      "learning_rate": 4.697314289659051e-05,
      "loss": 0.1265,
      "step": 333
    },
    {
      "epoch": 0.9355742296918768,
      "grad_norm": 0.5639265179634094,
      "learning_rate": 4.695139038269303e-05,
      "loss": 0.2573,
      "step": 334
    },
    {
      "epoch": 0.938375350140056,
      "grad_norm": 0.24481046199798584,
      "learning_rate": 4.6929565060064864e-05,
      "loss": 0.1265,
      "step": 335
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.2361876666545868,
      "learning_rate": 4.690766700109659e-05,
      "loss": 0.171,
      "step": 336
    },
    {
      "epoch": 0.9439775910364145,
      "grad_norm": 0.5476352572441101,
      "learning_rate": 4.688569627842007e-05,
      "loss": 0.2601,
      "step": 337
    },
    {
      "epoch": 0.9467787114845938,
      "grad_norm": 0.4084392786026001,
      "learning_rate": 4.686365296490813e-05,
      "loss": 0.0887,
      "step": 338
    },
    {
      "epoch": 0.9495798319327731,
      "grad_norm": 0.4765351414680481,
      "learning_rate": 4.684153713367442e-05,
      "loss": 0.2469,
      "step": 339
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.24255749583244324,
      "learning_rate": 4.681934885807308e-05,
      "loss": 0.1327,
      "step": 340
    },
    {
      "epoch": 0.9551820728291317,
      "grad_norm": 0.2432311475276947,
      "learning_rate": 4.6797088211698524e-05,
      "loss": 0.1349,
      "step": 341
    },
    {
      "epoch": 0.957983193277311,
      "grad_norm": 0.41482043266296387,
      "learning_rate": 4.677475526838526e-05,
      "loss": 0.0934,
      "step": 342
    },
    {
      "epoch": 0.9607843137254902,
      "grad_norm": 0.4962022006511688,
      "learning_rate": 4.675235010220754e-05,
      "loss": 0.2515,
      "step": 343
    },
    {
      "epoch": 0.9635854341736695,
      "grad_norm": 0.26970216631889343,
      "learning_rate": 4.672987278747919e-05,
      "loss": 0.2126,
      "step": 344
    },
    {
      "epoch": 0.9663865546218487,
      "grad_norm": 0.1891782432794571,
      "learning_rate": 4.6707323398753346e-05,
      "loss": 0.1656,
      "step": 345
    },
    {
      "epoch": 0.969187675070028,
      "grad_norm": 0.18780910968780518,
      "learning_rate": 4.668470201082218e-05,
      "loss": 0.1719,
      "step": 346
    },
    {
      "epoch": 0.9719887955182073,
      "grad_norm": 0.6354146599769592,
      "learning_rate": 4.6662008698716675e-05,
      "loss": 0.2906,
      "step": 347
    },
    {
      "epoch": 0.9747899159663865,
      "grad_norm": 0.19006231427192688,
      "learning_rate": 4.663924353770639e-05,
      "loss": 0.1676,
      "step": 348
    },
    {
      "epoch": 0.9775910364145658,
      "grad_norm": 0.1768893450498581,
      "learning_rate": 4.6616406603299176e-05,
      "loss": 0.1664,
      "step": 349
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 0.19590170681476593,
      "learning_rate": 4.6593497971240956e-05,
      "loss": 0.1647,
      "step": 350
    },
    {
      "epoch": 0.9831932773109243,
      "grad_norm": 0.4365139901638031,
      "learning_rate": 4.657051771751546e-05,
      "loss": 0.0991,
      "step": 351
    },
    {
      "epoch": 0.9859943977591037,
      "grad_norm": 0.7374207377433777,
      "learning_rate": 4.654746591834396e-05,
      "loss": 0.31,
      "step": 352
    },
    {
      "epoch": 0.988795518207283,
      "grad_norm": 0.2521171271800995,
      "learning_rate": 4.652434265018504e-05,
      "loss": 0.1356,
      "step": 353
    },
    {
      "epoch": 0.9915966386554622,
      "grad_norm": 0.2702823281288147,
      "learning_rate": 4.6501147989734346e-05,
      "loss": 0.2081,
      "step": 354
    },
    {
      "epoch": 0.9943977591036415,
      "grad_norm": 0.24237488210201263,
      "learning_rate": 4.647788201392429e-05,
      "loss": 0.175,
      "step": 355
    },
    {
      "epoch": 0.9971988795518207,
      "grad_norm": 0.2869884669780731,
      "learning_rate": 4.645454479992386e-05,
      "loss": 0.1384,
      "step": 356
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4562254548072815,
      "learning_rate": 4.64311364251383e-05,
      "loss": 0.1006,
      "step": 357
    },
    {
      "epoch": 1.0,
      "eval_f1 (minor class)": 0.12357723577235774,
      "eval_loss": 0.1737803816795349,
      "eval_roc_auc": 0.5386335507088105,
      "eval_runtime": 2.9217,
      "eval_samples_per_second": 433.989,
      "eval_steps_per_second": 13.69,
      "step": 357
    },
    {
      "epoch": 1.0028011204481793,
      "grad_norm": 0.2548682689666748,
      "learning_rate": 4.6407656967208876e-05,
      "loss": 0.2014,
      "step": 358
    },
    {
      "epoch": 1.0056022408963585,
      "grad_norm": 0.2920522689819336,
      "learning_rate": 4.638410650401267e-05,
      "loss": 0.144,
      "step": 359
    },
    {
      "epoch": 1.0084033613445378,
      "grad_norm": 0.44480934739112854,
      "learning_rate": 4.6360485113662216e-05,
      "loss": 0.0998,
      "step": 360
    },
    {
      "epoch": 1.011204481792717,
      "grad_norm": 0.44798406958580017,
      "learning_rate": 4.633679287450534e-05,
      "loss": 0.0989,
      "step": 361
    },
    {
      "epoch": 1.0140056022408963,
      "grad_norm": 0.21220609545707703,
      "learning_rate": 4.631302986512485e-05,
      "loss": 0.1732,
      "step": 362
    },
    {
      "epoch": 1.0168067226890756,
      "grad_norm": 0.4394678771495819,
      "learning_rate": 4.628919616433827e-05,
      "loss": 0.2458,
      "step": 363
    },
    {
      "epoch": 1.0196078431372548,
      "grad_norm": 0.6233344674110413,
      "learning_rate": 4.6265291851197626e-05,
      "loss": 0.2756,
      "step": 364
    },
    {
      "epoch": 1.022408963585434,
      "grad_norm": 0.43582528829574585,
      "learning_rate": 4.6241317004989126e-05,
      "loss": 0.2477,
      "step": 365
    },
    {
      "epoch": 1.0252100840336134,
      "grad_norm": 0.22646282613277435,
      "learning_rate": 4.621727170523293e-05,
      "loss": 0.1612,
      "step": 366
    },
    {
      "epoch": 1.0280112044817926,
      "grad_norm": 0.2966563403606415,
      "learning_rate": 4.619315603168289e-05,
      "loss": 0.1287,
      "step": 367
    },
    {
      "epoch": 1.0308123249299719,
      "grad_norm": 0.45250192284584045,
      "learning_rate": 4.6168970064326266e-05,
      "loss": 0.0975,
      "step": 368
    },
    {
      "epoch": 1.0336134453781514,
      "grad_norm": 1.0906392335891724,
      "learning_rate": 4.614471388338346e-05,
      "loss": 0.3621,
      "step": 369
    },
    {
      "epoch": 1.0364145658263306,
      "grad_norm": 0.3004130721092224,
      "learning_rate": 4.6120387569307775e-05,
      "loss": 0.1335,
      "step": 370
    },
    {
      "epoch": 1.0392156862745099,
      "grad_norm": 0.20164428651332855,
      "learning_rate": 4.609599120278514e-05,
      "loss": 0.1761,
      "step": 371
    },
    {
      "epoch": 1.0420168067226891,
      "grad_norm": 0.4512525796890259,
      "learning_rate": 4.6071524864733794e-05,
      "loss": 0.0978,
      "step": 372
    },
    {
      "epoch": 1.0448179271708684,
      "grad_norm": 0.2054992914199829,
      "learning_rate": 4.604698863630411e-05,
      "loss": 0.1603,
      "step": 373
    },
    {
      "epoch": 1.0476190476190477,
      "grad_norm": 0.27855244278907776,
      "learning_rate": 4.602238259887825e-05,
      "loss": 0.1278,
      "step": 374
    },
    {
      "epoch": 1.050420168067227,
      "grad_norm": 0.2679445743560791,
      "learning_rate": 4.599770683406991e-05,
      "loss": 0.1353,
      "step": 375
    },
    {
      "epoch": 1.0532212885154062,
      "grad_norm": 0.4758155941963196,
      "learning_rate": 4.5972961423724087e-05,
      "loss": 0.2309,
      "step": 376
    },
    {
      "epoch": 1.0560224089635855,
      "grad_norm": 0.4918065071105957,
      "learning_rate": 4.594814644991674e-05,
      "loss": 0.2486,
      "step": 377
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 0.45264533162117004,
      "learning_rate": 4.592326199495461e-05,
      "loss": 0.0964,
      "step": 378
    },
    {
      "epoch": 1.061624649859944,
      "grad_norm": 0.2610427439212799,
      "learning_rate": 4.5898308141374836e-05,
      "loss": 0.1328,
      "step": 379
    },
    {
      "epoch": 1.0644257703081232,
      "grad_norm": 0.6690388917922974,
      "learning_rate": 4.5873284971944784e-05,
      "loss": 0.2738,
      "step": 380
    },
    {
      "epoch": 1.0672268907563025,
      "grad_norm": 0.28492769598960876,
      "learning_rate": 4.5848192569661706e-05,
      "loss": 0.1286,
      "step": 381
    },
    {
      "epoch": 1.0700280112044818,
      "grad_norm": 0.30735185742378235,
      "learning_rate": 4.5823031017752485e-05,
      "loss": 0.2082,
      "step": 382
    },
    {
      "epoch": 1.072829131652661,
      "grad_norm": 0.390432208776474,
      "learning_rate": 4.579780039967339e-05,
      "loss": 0.2061,
      "step": 383
    },
    {
      "epoch": 1.0756302521008403,
      "grad_norm": 0.5083768367767334,
      "learning_rate": 4.577250079910973e-05,
      "loss": 0.2584,
      "step": 384
    },
    {
      "epoch": 1.0784313725490196,
      "grad_norm": 0.2757357060909271,
      "learning_rate": 4.574713229997563e-05,
      "loss": 0.1371,
      "step": 385
    },
    {
      "epoch": 1.0812324929971988,
      "grad_norm": 0.5543646216392517,
      "learning_rate": 4.5721694986413753e-05,
      "loss": 0.2586,
      "step": 386
    },
    {
      "epoch": 1.084033613445378,
      "grad_norm": 0.29792216420173645,
      "learning_rate": 4.5696188942795e-05,
      "loss": 0.21,
      "step": 387
    },
    {
      "epoch": 1.0868347338935573,
      "grad_norm": 0.2785909175872803,
      "learning_rate": 4.5670614253718224e-05,
      "loss": 0.213,
      "step": 388
    },
    {
      "epoch": 1.0896358543417366,
      "grad_norm": 0.6011048555374146,
      "learning_rate": 4.5644971004009984e-05,
      "loss": 0.2512,
      "step": 389
    },
    {
      "epoch": 1.092436974789916,
      "grad_norm": 0.32919472455978394,
      "learning_rate": 4.5619259278724214e-05,
      "loss": 0.1317,
      "step": 390
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 0.19824554026126862,
      "learning_rate": 4.5593479163141994e-05,
      "loss": 0.1699,
      "step": 391
    },
    {
      "epoch": 1.0980392156862746,
      "grad_norm": 0.20258326828479767,
      "learning_rate": 4.556763074277124e-05,
      "loss": 0.1732,
      "step": 392
    },
    {
      "epoch": 1.1008403361344539,
      "grad_norm": 0.30259111523628235,
      "learning_rate": 4.55417141033464e-05,
      "loss": 0.138,
      "step": 393
    },
    {
      "epoch": 1.1036414565826331,
      "grad_norm": 0.3482188880443573,
      "learning_rate": 4.551572933082822e-05,
      "loss": 0.1307,
      "step": 394
    },
    {
      "epoch": 1.1064425770308124,
      "grad_norm": 0.24273444712162018,
      "learning_rate": 4.548967651140341e-05,
      "loss": 0.2059,
      "step": 395
    },
    {
      "epoch": 1.1092436974789917,
      "grad_norm": 0.2121374011039734,
      "learning_rate": 4.54635557314844e-05,
      "loss": 0.1707,
      "step": 396
    },
    {
      "epoch": 1.112044817927171,
      "grad_norm": 0.22848618030548096,
      "learning_rate": 4.5437367077709e-05,
      "loss": 0.2048,
      "step": 397
    },
    {
      "epoch": 1.1148459383753502,
      "grad_norm": 0.4918600022792816,
      "learning_rate": 4.541111063694019e-05,
      "loss": 0.1049,
      "step": 398
    },
    {
      "epoch": 1.1176470588235294,
      "grad_norm": 0.47923439741134644,
      "learning_rate": 4.538478649626574e-05,
      "loss": 0.0996,
      "step": 399
    },
    {
      "epoch": 1.1204481792717087,
      "grad_norm": 0.42145538330078125,
      "learning_rate": 4.5358394742998e-05,
      "loss": 0.2455,
      "step": 400
    },
    {
      "epoch": 1.123249299719888,
      "grad_norm": 0.8408336639404297,
      "learning_rate": 4.533193546467357e-05,
      "loss": 0.3064,
      "step": 401
    },
    {
      "epoch": 1.1260504201680672,
      "grad_norm": 0.5717082619667053,
      "learning_rate": 4.530540874905302e-05,
      "loss": 0.2691,
      "step": 402
    },
    {
      "epoch": 1.1288515406162465,
      "grad_norm": 0.4880336821079254,
      "learning_rate": 4.527881468412058e-05,
      "loss": 0.1031,
      "step": 403
    },
    {
      "epoch": 1.1316526610644257,
      "grad_norm": 0.23222537338733673,
      "learning_rate": 4.52521533580839e-05,
      "loss": 0.2062,
      "step": 404
    },
    {
      "epoch": 1.134453781512605,
      "grad_norm": 0.4019043445587158,
      "learning_rate": 4.522542485937369e-05,
      "loss": 0.2336,
      "step": 405
    },
    {
      "epoch": 1.1372549019607843,
      "grad_norm": 0.4039878845214844,
      "learning_rate": 4.5198629276643465e-05,
      "loss": 0.2291,
      "step": 406
    },
    {
      "epoch": 1.1400560224089635,
      "grad_norm": 0.24419677257537842,
      "learning_rate": 4.517176669876927e-05,
      "loss": 0.1874,
      "step": 407
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.5417842268943787,
      "learning_rate": 4.5144837214849334e-05,
      "loss": 0.1077,
      "step": 408
    },
    {
      "epoch": 1.145658263305322,
      "grad_norm": 0.5518860816955566,
      "learning_rate": 4.5117840914203805e-05,
      "loss": 0.1073,
      "step": 409
    },
    {
      "epoch": 1.1484593837535013,
      "grad_norm": 0.282936692237854,
      "learning_rate": 4.509077788637446e-05,
      "loss": 0.1585,
      "step": 410
    },
    {
      "epoch": 1.1512605042016806,
      "grad_norm": 0.22835111618041992,
      "learning_rate": 4.5063648221124386e-05,
      "loss": 0.1841,
      "step": 411
    },
    {
      "epoch": 1.1540616246498598,
      "grad_norm": 0.36185768246650696,
      "learning_rate": 4.503645200843771e-05,
      "loss": 0.1392,
      "step": 412
    },
    {
      "epoch": 1.156862745098039,
      "grad_norm": 0.5351378321647644,
      "learning_rate": 4.500918933851928e-05,
      "loss": 0.1027,
      "step": 413
    },
    {
      "epoch": 1.1596638655462184,
      "grad_norm": 0.2676061987876892,
      "learning_rate": 4.498186030179434e-05,
      "loss": 0.158,
      "step": 414
    },
    {
      "epoch": 1.1624649859943978,
      "grad_norm": 0.3645527958869934,
      "learning_rate": 4.495446498890831e-05,
      "loss": 0.1314,
      "step": 415
    },
    {
      "epoch": 1.165266106442577,
      "grad_norm": 0.29105129837989807,
      "learning_rate": 4.4927003490726404e-05,
      "loss": 0.1341,
      "step": 416
    },
    {
      "epoch": 1.1680672268907564,
      "grad_norm": 0.7233947515487671,
      "learning_rate": 4.4899475898333367e-05,
      "loss": 0.2707,
      "step": 417
    },
    {
      "epoch": 1.1708683473389356,
      "grad_norm": 0.342913419008255,
      "learning_rate": 4.487188230303316e-05,
      "loss": 0.1763,
      "step": 418
    },
    {
      "epoch": 1.173669467787115,
      "grad_norm": 0.2501223385334015,
      "learning_rate": 4.48442227963487e-05,
      "loss": 0.1555,
      "step": 419
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.32546159625053406,
      "learning_rate": 4.4816497470021454e-05,
      "loss": 0.1288,
      "step": 420
    },
    {
      "epoch": 1.1792717086834734,
      "grad_norm": 0.23415644466876984,
      "learning_rate": 4.478870641601127e-05,
      "loss": 0.1575,
      "step": 421
    },
    {
      "epoch": 1.1820728291316527,
      "grad_norm": 0.2671819031238556,
      "learning_rate": 4.4760849726495945e-05,
      "loss": 0.13,
      "step": 422
    },
    {
      "epoch": 1.184873949579832,
      "grad_norm": 1.1153579950332642,
      "learning_rate": 4.473292749387102e-05,
      "loss": 0.3314,
      "step": 423
    },
    {
      "epoch": 1.1876750700280112,
      "grad_norm": 0.2628577947616577,
      "learning_rate": 4.47049398107494e-05,
      "loss": 0.1561,
      "step": 424
    },
    {
      "epoch": 1.1904761904761905,
      "grad_norm": 0.2286321073770523,
      "learning_rate": 4.467688676996111e-05,
      "loss": 0.1583,
      "step": 425
    },
    {
      "epoch": 1.1932773109243697,
      "grad_norm": 0.29673612117767334,
      "learning_rate": 4.464876846455291e-05,
      "loss": 0.1242,
      "step": 426
    },
    {
      "epoch": 1.196078431372549,
      "grad_norm": 0.44082918763160706,
      "learning_rate": 4.4620584987788065e-05,
      "loss": 0.2101,
      "step": 427
    },
    {
      "epoch": 1.1988795518207283,
      "grad_norm": 0.2973793148994446,
      "learning_rate": 4.4592336433146e-05,
      "loss": 0.1493,
      "step": 428
    },
    {
      "epoch": 1.2016806722689075,
      "grad_norm": 0.2873530089855194,
      "learning_rate": 4.4564022894321966e-05,
      "loss": 0.1652,
      "step": 429
    },
    {
      "epoch": 1.2044817927170868,
      "grad_norm": 0.5498677492141724,
      "learning_rate": 4.4535644465226796e-05,
      "loss": 0.0821,
      "step": 430
    },
    {
      "epoch": 1.207282913165266,
      "grad_norm": 0.249070942401886,
      "learning_rate": 4.450720123998651e-05,
      "loss": 0.1611,
      "step": 431
    },
    {
      "epoch": 1.2100840336134453,
      "grad_norm": 0.5754450559616089,
      "learning_rate": 4.4478693312942054e-05,
      "loss": 0.236,
      "step": 432
    },
    {
      "epoch": 1.2128851540616246,
      "grad_norm": 0.31797465682029724,
      "learning_rate": 4.4450120778649014e-05,
      "loss": 0.1199,
      "step": 433
    },
    {
      "epoch": 1.215686274509804,
      "grad_norm": 0.4542624056339264,
      "learning_rate": 4.4421483731877214e-05,
      "loss": 0.1787,
      "step": 434
    },
    {
      "epoch": 1.2184873949579833,
      "grad_norm": 0.35622158646583557,
      "learning_rate": 4.43927822676105e-05,
      "loss": 0.1261,
      "step": 435
    },
    {
      "epoch": 1.2212885154061626,
      "grad_norm": 0.9345407485961914,
      "learning_rate": 4.4364016481046336e-05,
      "loss": 0.2648,
      "step": 436
    },
    {
      "epoch": 1.2240896358543418,
      "grad_norm": 0.4623335301876068,
      "learning_rate": 4.433518646759558e-05,
      "loss": 0.187,
      "step": 437
    },
    {
      "epoch": 1.226890756302521,
      "grad_norm": 0.8016291856765747,
      "learning_rate": 4.4306292322882066e-05,
      "loss": 0.2376,
      "step": 438
    },
    {
      "epoch": 1.2296918767507004,
      "grad_norm": 0.3608405292034149,
      "learning_rate": 4.4277334142742375e-05,
      "loss": 0.1337,
      "step": 439
    },
    {
      "epoch": 1.2324929971988796,
      "grad_norm": 0.9408867359161377,
      "learning_rate": 4.424831202322548e-05,
      "loss": 0.2775,
      "step": 440
    },
    {
      "epoch": 1.2352941176470589,
      "grad_norm": 0.36630916595458984,
      "learning_rate": 4.421922606059242e-05,
      "loss": 0.1437,
      "step": 441
    },
    {
      "epoch": 1.2380952380952381,
      "grad_norm": 0.5501814484596252,
      "learning_rate": 4.419007635131598e-05,
      "loss": 0.129,
      "step": 442
    },
    {
      "epoch": 1.2408963585434174,
      "grad_norm": 0.5585311651229858,
      "learning_rate": 4.416086299208041e-05,
      "loss": 0.2171,
      "step": 443
    },
    {
      "epoch": 1.2436974789915967,
      "grad_norm": 0.5523871779441833,
      "learning_rate": 4.413158607978104e-05,
      "loss": 0.1953,
      "step": 444
    },
    {
      "epoch": 1.246498599439776,
      "grad_norm": 0.778249979019165,
      "learning_rate": 4.410224571152403e-05,
      "loss": 0.0988,
      "step": 445
    },
    {
      "epoch": 1.2492997198879552,
      "grad_norm": 0.3328462839126587,
      "learning_rate": 4.407284198462597e-05,
      "loss": 0.189,
      "step": 446
    },
    {
      "epoch": 1.2521008403361344,
      "grad_norm": 0.3411165177822113,
      "learning_rate": 4.404337499661364e-05,
      "loss": 0.1717,
      "step": 447
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 0.3889773488044739,
      "learning_rate": 4.4013844845223626e-05,
      "loss": 0.2114,
      "step": 448
    },
    {
      "epoch": 1.257703081232493,
      "grad_norm": 0.2957543134689331,
      "learning_rate": 4.398425162840202e-05,
      "loss": 0.1762,
      "step": 449
    },
    {
      "epoch": 1.2605042016806722,
      "grad_norm": 0.4489629864692688,
      "learning_rate": 4.395459544430407e-05,
      "loss": 0.2177,
      "step": 450
    },
    {
      "epoch": 1.2633053221288515,
      "grad_norm": 0.5136566162109375,
      "learning_rate": 4.3924876391293915e-05,
      "loss": 0.1395,
      "step": 451
    },
    {
      "epoch": 1.2661064425770308,
      "grad_norm": 0.6125023365020752,
      "learning_rate": 4.3895094567944186e-05,
      "loss": 0.1241,
      "step": 452
    },
    {
      "epoch": 1.26890756302521,
      "grad_norm": 0.3510366380214691,
      "learning_rate": 4.386525007303571e-05,
      "loss": 0.1614,
      "step": 453
    },
    {
      "epoch": 1.2717086834733893,
      "grad_norm": 0.31088364124298096,
      "learning_rate": 4.3835343005557215e-05,
      "loss": 0.1631,
      "step": 454
    },
    {
      "epoch": 1.2745098039215685,
      "grad_norm": 0.4494378864765167,
      "learning_rate": 4.380537346470495e-05,
      "loss": 0.1324,
      "step": 455
    },
    {
      "epoch": 1.2773109243697478,
      "grad_norm": 0.5657458901405334,
      "learning_rate": 4.3775341549882364e-05,
      "loss": 0.1197,
      "step": 456
    },
    {
      "epoch": 1.280112044817927,
      "grad_norm": 0.26143553853034973,
      "learning_rate": 4.374524736069982e-05,
      "loss": 0.1596,
      "step": 457
    },
    {
      "epoch": 1.2829131652661063,
      "grad_norm": 0.3509024977684021,
      "learning_rate": 4.37150909969742e-05,
      "loss": 0.1464,
      "step": 458
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.3572234809398651,
      "learning_rate": 4.368487255872864e-05,
      "loss": 0.1751,
      "step": 459
    },
    {
      "epoch": 1.2885154061624648,
      "grad_norm": 0.50468909740448,
      "learning_rate": 4.365459214619214e-05,
      "loss": 0.0963,
      "step": 460
    },
    {
      "epoch": 1.2913165266106443,
      "grad_norm": 0.3895762860774994,
      "learning_rate": 4.3624249859799274e-05,
      "loss": 0.1136,
      "step": 461
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 0.6248763203620911,
      "learning_rate": 4.359384580018982e-05,
      "loss": 0.1675,
      "step": 462
    },
    {
      "epoch": 1.2969187675070029,
      "grad_norm": 0.37576088309288025,
      "learning_rate": 4.356338006820849e-05,
      "loss": 0.1191,
      "step": 463
    },
    {
      "epoch": 1.2997198879551821,
      "grad_norm": 0.5103853940963745,
      "learning_rate": 4.35328527649045e-05,
      "loss": 0.1591,
      "step": 464
    },
    {
      "epoch": 1.3025210084033614,
      "grad_norm": 0.6339231133460999,
      "learning_rate": 4.35022639915313e-05,
      "loss": 0.1657,
      "step": 465
    },
    {
      "epoch": 1.3053221288515406,
      "grad_norm": 0.3810737729072571,
      "learning_rate": 4.347161384954626e-05,
      "loss": 0.1179,
      "step": 466
    },
    {
      "epoch": 1.30812324929972,
      "grad_norm": 1.3601053953170776,
      "learning_rate": 4.344090244061024e-05,
      "loss": 0.2517,
      "step": 467
    },
    {
      "epoch": 1.3109243697478992,
      "grad_norm": 0.40818560123443604,
      "learning_rate": 4.341012986658738e-05,
      "loss": 0.123,
      "step": 468
    },
    {
      "epoch": 1.3137254901960784,
      "grad_norm": 1.1468818187713623,
      "learning_rate": 4.337929622954463e-05,
      "loss": 0.2502,
      "step": 469
    },
    {
      "epoch": 1.3165266106442577,
      "grad_norm": 1.0096850395202637,
      "learning_rate": 4.334840163175151e-05,
      "loss": 0.2248,
      "step": 470
    },
    {
      "epoch": 1.319327731092437,
      "grad_norm": 0.43895837664604187,
      "learning_rate": 4.3317446175679735e-05,
      "loss": 0.1442,
      "step": 471
    },
    {
      "epoch": 1.3221288515406162,
      "grad_norm": 1.211270809173584,
      "learning_rate": 4.3286429964002856e-05,
      "loss": 0.2676,
      "step": 472
    },
    {
      "epoch": 1.3249299719887955,
      "grad_norm": 0.5929017663002014,
      "learning_rate": 4.325535309959596e-05,
      "loss": 0.0812,
      "step": 473
    },
    {
      "epoch": 1.3277310924369747,
      "grad_norm": 0.3913727104663849,
      "learning_rate": 4.3224215685535294e-05,
      "loss": 0.1334,
      "step": 474
    },
    {
      "epoch": 1.330532212885154,
      "grad_norm": 0.4079723358154297,
      "learning_rate": 4.3193017825097936e-05,
      "loss": 0.1742,
      "step": 475
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.3363523781299591,
      "learning_rate": 4.316175962176148e-05,
      "loss": 0.1315,
      "step": 476
    },
    {
      "epoch": 1.3361344537815127,
      "grad_norm": 0.5548957586288452,
      "learning_rate": 4.313044117920363e-05,
      "loss": 0.0904,
      "step": 477
    },
    {
      "epoch": 1.338935574229692,
      "grad_norm": 0.2717703580856323,
      "learning_rate": 4.3099062601301904e-05,
      "loss": 0.1617,
      "step": 478
    },
    {
      "epoch": 1.3417366946778713,
      "grad_norm": 0.48116743564605713,
      "learning_rate": 4.30676239921333e-05,
      "loss": 0.2158,
      "step": 479
    },
    {
      "epoch": 1.3445378151260505,
      "grad_norm": 0.5152187943458557,
      "learning_rate": 4.3036125455973896e-05,
      "loss": 0.0882,
      "step": 480
    },
    {
      "epoch": 1.3473389355742298,
      "grad_norm": 0.747715175151825,
      "learning_rate": 4.300456709729856e-05,
      "loss": 0.2583,
      "step": 481
    },
    {
      "epoch": 1.350140056022409,
      "grad_norm": 0.24533413350582123,
      "learning_rate": 4.2972949020780575e-05,
      "loss": 0.1618,
      "step": 482
    },
    {
      "epoch": 1.3529411764705883,
      "grad_norm": 0.2675352394580841,
      "learning_rate": 4.294127133129128e-05,
      "loss": 0.1735,
      "step": 483
    },
    {
      "epoch": 1.3557422969187676,
      "grad_norm": 0.40604180097579956,
      "learning_rate": 4.290953413389977e-05,
      "loss": 0.1971,
      "step": 484
    },
    {
      "epoch": 1.3585434173669468,
      "grad_norm": 0.5014567971229553,
      "learning_rate": 4.2877737533872485e-05,
      "loss": 0.0903,
      "step": 485
    },
    {
      "epoch": 1.361344537815126,
      "grad_norm": 0.3054341971874237,
      "learning_rate": 4.284588163667292e-05,
      "loss": 0.1307,
      "step": 486
    },
    {
      "epoch": 1.3641456582633054,
      "grad_norm": 0.5000020265579224,
      "learning_rate": 4.281396654796124e-05,
      "loss": 0.0904,
      "step": 487
    },
    {
      "epoch": 1.3669467787114846,
      "grad_norm": 0.3115267753601074,
      "learning_rate": 4.278199237359392e-05,
      "loss": 0.1209,
      "step": 488
    },
    {
      "epoch": 1.3697478991596639,
      "grad_norm": 0.22438497841358185,
      "learning_rate": 4.274995921962343e-05,
      "loss": 0.1775,
      "step": 489
    },
    {
      "epoch": 1.3725490196078431,
      "grad_norm": 0.25052976608276367,
      "learning_rate": 4.271786719229786e-05,
      "loss": 0.1337,
      "step": 490
    },
    {
      "epoch": 1.3753501400560224,
      "grad_norm": 0.2692072093486786,
      "learning_rate": 4.268571639806057e-05,
      "loss": 0.1773,
      "step": 491
    },
    {
      "epoch": 1.3781512605042017,
      "grad_norm": 0.5472849011421204,
      "learning_rate": 4.265350694354985e-05,
      "loss": 0.2203,
      "step": 492
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 0.2849833071231842,
      "learning_rate": 4.2621238935598524e-05,
      "loss": 0.1256,
      "step": 493
    },
    {
      "epoch": 1.3837535014005602,
      "grad_norm": 0.4663335084915161,
      "learning_rate": 4.2588912481233666e-05,
      "loss": 0.0828,
      "step": 494
    },
    {
      "epoch": 1.3865546218487395,
      "grad_norm": 0.3137625753879547,
      "learning_rate": 4.2556527687676186e-05,
      "loss": 0.1409,
      "step": 495
    },
    {
      "epoch": 1.3893557422969187,
      "grad_norm": 0.23330673575401306,
      "learning_rate": 4.2524084662340494e-05,
      "loss": 0.1695,
      "step": 496
    },
    {
      "epoch": 1.392156862745098,
      "grad_norm": 0.2527044117450714,
      "learning_rate": 4.249158351283414e-05,
      "loss": 0.125,
      "step": 497
    },
    {
      "epoch": 1.3949579831932772,
      "grad_norm": 0.2422427237033844,
      "learning_rate": 4.2459024346957475e-05,
      "loss": 0.1719,
      "step": 498
    },
    {
      "epoch": 1.3977591036414565,
      "grad_norm": 0.25663718581199646,
      "learning_rate": 4.2426407272703284e-05,
      "loss": 0.1757,
      "step": 499
    },
    {
      "epoch": 1.4005602240896358,
      "grad_norm": 0.2737230062484741,
      "learning_rate": 4.2393732398256394e-05,
      "loss": 0.114,
      "step": 500
    },
    {
      "epoch": 1.403361344537815,
      "grad_norm": 0.2504082918167114,
      "learning_rate": 4.236099983199338e-05,
      "loss": 0.1697,
      "step": 501
    },
    {
      "epoch": 1.4061624649859943,
      "grad_norm": 0.911520779132843,
      "learning_rate": 4.232820968248214e-05,
      "loss": 0.2774,
      "step": 502
    },
    {
      "epoch": 1.4089635854341735,
      "grad_norm": 0.4278314411640167,
      "learning_rate": 4.229536205848158e-05,
      "loss": 0.2105,
      "step": 503
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 0.22557204961776733,
      "learning_rate": 4.2262457068941245e-05,
      "loss": 0.1257,
      "step": 504
    },
    {
      "epoch": 1.4145658263305323,
      "grad_norm": 0.2534330189228058,
      "learning_rate": 4.222949482300094e-05,
      "loss": 0.1229,
      "step": 505
    },
    {
      "epoch": 1.4173669467787116,
      "grad_norm": 0.20387370884418488,
      "learning_rate": 4.219647542999037e-05,
      "loss": 0.1262,
      "step": 506
    },
    {
      "epoch": 1.4201680672268908,
      "grad_norm": 0.20558597147464752,
      "learning_rate": 4.21633989994288e-05,
      "loss": 0.1239,
      "step": 507
    },
    {
      "epoch": 1.42296918767507,
      "grad_norm": 0.6656189560890198,
      "learning_rate": 4.2130265641024705e-05,
      "loss": 0.2589,
      "step": 508
    },
    {
      "epoch": 1.4257703081232493,
      "grad_norm": 0.3030642569065094,
      "learning_rate": 4.209707546467531e-05,
      "loss": 0.1137,
      "step": 509
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.2517474591732025,
      "learning_rate": 4.206382858046636e-05,
      "loss": 0.1739,
      "step": 510
    },
    {
      "epoch": 1.4313725490196079,
      "grad_norm": 0.21296930313110352,
      "learning_rate": 4.2030525098671646e-05,
      "loss": 0.1274,
      "step": 511
    },
    {
      "epoch": 1.4341736694677871,
      "grad_norm": 0.23091089725494385,
      "learning_rate": 4.199716512975272e-05,
      "loss": 0.1305,
      "step": 512
    },
    {
      "epoch": 1.4369747899159664,
      "grad_norm": 0.25984320044517517,
      "learning_rate": 4.1963748784358456e-05,
      "loss": 0.1668,
      "step": 513
    },
    {
      "epoch": 1.4397759103641457,
      "grad_norm": 0.2812284231185913,
      "learning_rate": 4.1930276173324756e-05,
      "loss": 0.1218,
      "step": 514
    },
    {
      "epoch": 1.442577030812325,
      "grad_norm": 0.42426514625549316,
      "learning_rate": 4.189674740767411e-05,
      "loss": 0.0794,
      "step": 515
    },
    {
      "epoch": 1.4453781512605042,
      "grad_norm": 0.23902854323387146,
      "learning_rate": 4.1863162598615265e-05,
      "loss": 0.1354,
      "step": 516
    },
    {
      "epoch": 1.4481792717086834,
      "grad_norm": 0.2747328281402588,
      "learning_rate": 4.1829521857542885e-05,
      "loss": 0.1168,
      "step": 517
    },
    {
      "epoch": 1.4509803921568627,
      "grad_norm": 0.25116053223609924,
      "learning_rate": 4.1795825296037126e-05,
      "loss": 0.1312,
      "step": 518
    },
    {
      "epoch": 1.453781512605042,
      "grad_norm": 0.7276238799095154,
      "learning_rate": 4.176207302586329e-05,
      "loss": 0.2773,
      "step": 519
    },
    {
      "epoch": 1.4565826330532212,
      "grad_norm": 0.26795586943626404,
      "learning_rate": 4.172826515897146e-05,
      "loss": 0.1678,
      "step": 520
    },
    {
      "epoch": 1.4593837535014005,
      "grad_norm": 0.42587825655937195,
      "learning_rate": 4.169440180749612e-05,
      "loss": 0.0729,
      "step": 521
    },
    {
      "epoch": 1.46218487394958,
      "grad_norm": 0.3999551832675934,
      "learning_rate": 4.166048308375579e-05,
      "loss": 0.1851,
      "step": 522
    },
    {
      "epoch": 1.4649859943977592,
      "grad_norm": 0.2874641418457031,
      "learning_rate": 4.162650910025264e-05,
      "loss": 0.162,
      "step": 523
    },
    {
      "epoch": 1.4677871148459385,
      "grad_norm": 0.7796664834022522,
      "learning_rate": 4.159247996967215e-05,
      "loss": 0.2352,
      "step": 524
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 1.0714823007583618,
      "learning_rate": 4.1558395804882695e-05,
      "loss": 0.3158,
      "step": 525
    },
    {
      "epoch": 1.473389355742297,
      "grad_norm": 1.209743618965149,
      "learning_rate": 4.152425671893518e-05,
      "loss": 0.3378,
      "step": 526
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 0.25424477458000183,
      "learning_rate": 4.149006282506268e-05,
      "loss": 0.1739,
      "step": 527
    },
    {
      "epoch": 1.4789915966386555,
      "grad_norm": 0.2382468283176422,
      "learning_rate": 4.145581423668008e-05,
      "loss": 0.1301,
      "step": 528
    },
    {
      "epoch": 1.4817927170868348,
      "grad_norm": 0.2188778817653656,
      "learning_rate": 4.142151106738364e-05,
      "loss": 0.1625,
      "step": 529
    },
    {
      "epoch": 1.484593837535014,
      "grad_norm": 0.4427192807197571,
      "learning_rate": 4.138715343095068e-05,
      "loss": 0.085,
      "step": 530
    },
    {
      "epoch": 1.4873949579831933,
      "grad_norm": 0.22112633287906647,
      "learning_rate": 4.135274144133918e-05,
      "loss": 0.1694,
      "step": 531
    },
    {
      "epoch": 1.4901960784313726,
      "grad_norm": 0.27271297574043274,
      "learning_rate": 4.1318275212687376e-05,
      "loss": 0.1251,
      "step": 532
    },
    {
      "epoch": 1.4929971988795518,
      "grad_norm": 0.3934704661369324,
      "learning_rate": 4.1283754859313414e-05,
      "loss": 0.2116,
      "step": 533
    },
    {
      "epoch": 1.495798319327731,
      "grad_norm": 0.33457499742507935,
      "learning_rate": 4.124918049571499e-05,
      "loss": 0.2102,
      "step": 534
    },
    {
      "epoch": 1.4985994397759104,
      "grad_norm": 0.33014121651649475,
      "learning_rate": 4.12145522365689e-05,
      "loss": 0.2068,
      "step": 535
    },
    {
      "epoch": 1.5014005602240896,
      "grad_norm": 1.0461491346359253,
      "learning_rate": 4.117987019673073e-05,
      "loss": 0.3212,
      "step": 536
    },
    {
      "epoch": 1.504201680672269,
      "grad_norm": 0.45663779973983765,
      "learning_rate": 4.1145134491234427e-05,
      "loss": 0.2363,
      "step": 537
    },
    {
      "epoch": 1.5070028011204482,
      "grad_norm": 0.46917828917503357,
      "learning_rate": 4.111034523529196e-05,
      "loss": 0.2326,
      "step": 538
    },
    {
      "epoch": 1.5098039215686274,
      "grad_norm": 0.30291152000427246,
      "learning_rate": 4.1075502544292884e-05,
      "loss": 0.2064,
      "step": 539
    },
    {
      "epoch": 1.5126050420168067,
      "grad_norm": 0.3694068491458893,
      "learning_rate": 4.1040606533804024e-05,
      "loss": 0.1405,
      "step": 540
    },
    {
      "epoch": 1.515406162464986,
      "grad_norm": 0.3829762935638428,
      "learning_rate": 4.100565731956903e-05,
      "loss": 0.1385,
      "step": 541
    },
    {
      "epoch": 1.5182072829131652,
      "grad_norm": 0.2339664101600647,
      "learning_rate": 4.097065501750804e-05,
      "loss": 0.1658,
      "step": 542
    },
    {
      "epoch": 1.5210084033613445,
      "grad_norm": 0.2567518353462219,
      "learning_rate": 4.093559974371725e-05,
      "loss": 0.1748,
      "step": 543
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 0.5793279409408569,
      "learning_rate": 4.0900491614468553e-05,
      "loss": 0.1087,
      "step": 544
    },
    {
      "epoch": 1.526610644257703,
      "grad_norm": 0.5785588622093201,
      "learning_rate": 4.086533074620919e-05,
      "loss": 0.1099,
      "step": 545
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 0.5617150664329529,
      "learning_rate": 4.083011725556129e-05,
      "loss": 0.1074,
      "step": 546
    },
    {
      "epoch": 1.5322128851540615,
      "grad_norm": 0.36825308203697205,
      "learning_rate": 4.0794851259321546e-05,
      "loss": 0.2374,
      "step": 547
    },
    {
      "epoch": 1.5350140056022408,
      "grad_norm": 0.5937955379486084,
      "learning_rate": 4.0759532874460785e-05,
      "loss": 0.2677,
      "step": 548
    },
    {
      "epoch": 1.53781512605042,
      "grad_norm": 0.24256612360477448,
      "learning_rate": 4.0724162218123596e-05,
      "loss": 0.1698,
      "step": 549
    },
    {
      "epoch": 1.5406162464985993,
      "grad_norm": 0.5699710249900818,
      "learning_rate": 4.068873940762796e-05,
      "loss": 0.1086,
      "step": 550
    },
    {
      "epoch": 1.5434173669467786,
      "grad_norm": 0.3110926151275635,
      "learning_rate": 4.065326456046483e-05,
      "loss": 0.2061,
      "step": 551
    },
    {
      "epoch": 1.5462184873949578,
      "grad_norm": 0.36440059542655945,
      "learning_rate": 4.0617737794297764e-05,
      "loss": 0.2146,
      "step": 552
    },
    {
      "epoch": 1.5490196078431373,
      "grad_norm": 0.2662959098815918,
      "learning_rate": 4.058215922696252e-05,
      "loss": 0.1728,
      "step": 553
    },
    {
      "epoch": 1.5518207282913166,
      "grad_norm": 0.3178268074989319,
      "learning_rate": 4.0546528976466655e-05,
      "loss": 0.1402,
      "step": 554
    },
    {
      "epoch": 1.5546218487394958,
      "grad_norm": 0.30780094861984253,
      "learning_rate": 4.051084716098921e-05,
      "loss": 0.2053,
      "step": 555
    },
    {
      "epoch": 1.557422969187675,
      "grad_norm": 0.21479925513267517,
      "learning_rate": 4.047511389888017e-05,
      "loss": 0.1748,
      "step": 556
    },
    {
      "epoch": 1.5602240896358543,
      "grad_norm": 0.24510297179222107,
      "learning_rate": 4.043932930866021e-05,
      "loss": 0.1683,
      "step": 557
    },
    {
      "epoch": 1.5630252100840336,
      "grad_norm": 0.2353736162185669,
      "learning_rate": 4.040349350902028e-05,
      "loss": 0.1643,
      "step": 558
    },
    {
      "epoch": 1.5658263305322129,
      "grad_norm": 0.5024756789207458,
      "learning_rate": 4.036760661882109e-05,
      "loss": 0.0958,
      "step": 559
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 0.20932908356189728,
      "learning_rate": 4.033166875709291e-05,
      "loss": 0.1738,
      "step": 560
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.20001398026943207,
      "learning_rate": 4.029568004303501e-05,
      "loss": 0.173,
      "step": 561
    },
    {
      "epoch": 1.5742296918767507,
      "grad_norm": 0.2733915150165558,
      "learning_rate": 4.025964059601535e-05,
      "loss": 0.1329,
      "step": 562
    },
    {
      "epoch": 1.57703081232493,
      "grad_norm": 0.48111703991889954,
      "learning_rate": 4.022355053557015e-05,
      "loss": 0.2336,
      "step": 563
    },
    {
      "epoch": 1.5798319327731094,
      "grad_norm": 0.6872272491455078,
      "learning_rate": 4.018740998140352e-05,
      "loss": 0.2748,
      "step": 564
    },
    {
      "epoch": 1.5826330532212887,
      "grad_norm": 0.7389792203903198,
      "learning_rate": 4.015121905338704e-05,
      "loss": 0.2948,
      "step": 565
    },
    {
      "epoch": 1.585434173669468,
      "grad_norm": 0.46626532077789307,
      "learning_rate": 4.011497787155938e-05,
      "loss": 0.2341,
      "step": 566
    },
    {
      "epoch": 1.5882352941176472,
      "grad_norm": 0.2073425054550171,
      "learning_rate": 4.007868655612586e-05,
      "loss": 0.1662,
      "step": 567
    },
    {
      "epoch": 1.5910364145658265,
      "grad_norm": 0.2893884479999542,
      "learning_rate": 4.004234522745813e-05,
      "loss": 0.1419,
      "step": 568
    },
    {
      "epoch": 1.5938375350140057,
      "grad_norm": 0.3202155828475952,
      "learning_rate": 4.00059540060937e-05,
      "loss": 0.1377,
      "step": 569
    },
    {
      "epoch": 1.596638655462185,
      "grad_norm": 0.21610812842845917,
      "learning_rate": 3.996951301273557e-05,
      "loss": 0.1696,
      "step": 570
    },
    {
      "epoch": 1.5994397759103642,
      "grad_norm": 0.1836165338754654,
      "learning_rate": 3.993302236825181e-05,
      "loss": 0.1733,
      "step": 571
    },
    {
      "epoch": 1.6022408963585435,
      "grad_norm": 0.2764729857444763,
      "learning_rate": 3.98964821936752e-05,
      "loss": 0.1683,
      "step": 572
    },
    {
      "epoch": 1.6050420168067228,
      "grad_norm": 0.31198838353157043,
      "learning_rate": 3.9859892610202785e-05,
      "loss": 0.1573,
      "step": 573
    },
    {
      "epoch": 1.607843137254902,
      "grad_norm": 0.5770326852798462,
      "learning_rate": 3.982325373919549e-05,
      "loss": 0.2388,
      "step": 574
    },
    {
      "epoch": 1.6106442577030813,
      "grad_norm": 0.3288671672344208,
      "learning_rate": 3.9786565702177723e-05,
      "loss": 0.2145,
      "step": 575
    },
    {
      "epoch": 1.6134453781512605,
      "grad_norm": 0.2905415892601013,
      "learning_rate": 3.974982862083697e-05,
      "loss": 0.1867,
      "step": 576
    },
    {
      "epoch": 1.6162464985994398,
      "grad_norm": 0.2127629816532135,
      "learning_rate": 3.9713042617023386e-05,
      "loss": 0.1782,
      "step": 577
    },
    {
      "epoch": 1.619047619047619,
      "grad_norm": 0.2230168879032135,
      "learning_rate": 3.967620781274938e-05,
      "loss": 0.1759,
      "step": 578
    },
    {
      "epoch": 1.6218487394957983,
      "grad_norm": 0.3231712281703949,
      "learning_rate": 3.9639324330189236e-05,
      "loss": 0.207,
      "step": 579
    },
    {
      "epoch": 1.6246498599439776,
      "grad_norm": 0.3769891858100891,
      "learning_rate": 3.960239229167869e-05,
      "loss": 0.1371,
      "step": 580
    },
    {
      "epoch": 1.6274509803921569,
      "grad_norm": 0.32875877618789673,
      "learning_rate": 3.956541181971455e-05,
      "loss": 0.1396,
      "step": 581
    },
    {
      "epoch": 1.6302521008403361,
      "grad_norm": 0.2713623046875,
      "learning_rate": 3.9528383036954224e-05,
      "loss": 0.1653,
      "step": 582
    },
    {
      "epoch": 1.6330532212885154,
      "grad_norm": 1.0410537719726562,
      "learning_rate": 3.949130606621541e-05,
      "loss": 0.3346,
      "step": 583
    },
    {
      "epoch": 1.6358543417366946,
      "grad_norm": 0.23961792886257172,
      "learning_rate": 3.945418103047558e-05,
      "loss": 0.1784,
      "step": 584
    },
    {
      "epoch": 1.638655462184874,
      "grad_norm": 0.286924809217453,
      "learning_rate": 3.941700805287168e-05,
      "loss": 0.2005,
      "step": 585
    },
    {
      "epoch": 1.6414565826330532,
      "grad_norm": 0.38071775436401367,
      "learning_rate": 3.937978725669965e-05,
      "loss": 0.1445,
      "step": 586
    },
    {
      "epoch": 1.6442577030812324,
      "grad_norm": 0.5587998032569885,
      "learning_rate": 3.934251876541404e-05,
      "loss": 0.1045,
      "step": 587
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 0.3966423273086548,
      "learning_rate": 3.9305202702627575e-05,
      "loss": 0.1404,
      "step": 588
    },
    {
      "epoch": 1.649859943977591,
      "grad_norm": 0.3944240212440491,
      "learning_rate": 3.92678391921108e-05,
      "loss": 0.139,
      "step": 589
    },
    {
      "epoch": 1.6526610644257702,
      "grad_norm": 0.25802481174468994,
      "learning_rate": 3.9230428357791595e-05,
      "loss": 0.1688,
      "step": 590
    },
    {
      "epoch": 1.6554621848739495,
      "grad_norm": 0.22108642756938934,
      "learning_rate": 3.919297032375485e-05,
      "loss": 0.1737,
      "step": 591
    },
    {
      "epoch": 1.6582633053221287,
      "grad_norm": 0.3680080473423004,
      "learning_rate": 3.915546521424198e-05,
      "loss": 0.1375,
      "step": 592
    },
    {
      "epoch": 1.661064425770308,
      "grad_norm": 0.2191331535577774,
      "learning_rate": 3.9117913153650546e-05,
      "loss": 0.1674,
      "step": 593
    },
    {
      "epoch": 1.6638655462184873,
      "grad_norm": 0.4968366324901581,
      "learning_rate": 3.908031426653383e-05,
      "loss": 0.0956,
      "step": 594
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.2787998616695404,
      "learning_rate": 3.9042668677600436e-05,
      "loss": 0.1356,
      "step": 595
    },
    {
      "epoch": 1.6694677871148458,
      "grad_norm": 0.37599462270736694,
      "learning_rate": 3.900497651171388e-05,
      "loss": 0.2054,
      "step": 596
    },
    {
      "epoch": 1.6722689075630253,
      "grad_norm": 0.28965288400650024,
      "learning_rate": 3.8967237893892134e-05,
      "loss": 0.1325,
      "step": 597
    },
    {
      "epoch": 1.6750700280112045,
      "grad_norm": 0.44063901901245117,
      "learning_rate": 3.892945294930728e-05,
      "loss": 0.0864,
      "step": 598
    },
    {
      "epoch": 1.6778711484593838,
      "grad_norm": 0.29227393865585327,
      "learning_rate": 3.889162180328505e-05,
      "loss": 0.1258,
      "step": 599
    },
    {
      "epoch": 1.680672268907563,
      "grad_norm": 0.3660801351070404,
      "learning_rate": 3.885374458130438e-05,
      "loss": 0.2033,
      "step": 600
    },
    {
      "epoch": 1.6834733893557423,
      "grad_norm": 0.4446983337402344,
      "learning_rate": 3.881582140899707e-05,
      "loss": 0.0834,
      "step": 601
    },
    {
      "epoch": 1.6862745098039216,
      "grad_norm": 0.7357985377311707,
      "learning_rate": 3.877785241214733e-05,
      "loss": 0.244,
      "step": 602
    },
    {
      "epoch": 1.6890756302521008,
      "grad_norm": 0.2903584837913513,
      "learning_rate": 3.873983771669133e-05,
      "loss": 0.1743,
      "step": 603
    },
    {
      "epoch": 1.69187675070028,
      "grad_norm": 0.2409980297088623,
      "learning_rate": 3.8701777448716856e-05,
      "loss": 0.1273,
      "step": 604
    },
    {
      "epoch": 1.6946778711484594,
      "grad_norm": 0.22482717037200928,
      "learning_rate": 3.866367173446281e-05,
      "loss": 0.1268,
      "step": 605
    },
    {
      "epoch": 1.6974789915966386,
      "grad_norm": 0.2362489402294159,
      "learning_rate": 3.862552070031886e-05,
      "loss": 0.1251,
      "step": 606
    },
    {
      "epoch": 1.7002801120448179,
      "grad_norm": 0.23961283266544342,
      "learning_rate": 3.858732447282497e-05,
      "loss": 0.1698,
      "step": 607
    },
    {
      "epoch": 1.7030812324929971,
      "grad_norm": 0.5947995185852051,
      "learning_rate": 3.854908317867102e-05,
      "loss": 0.2085,
      "step": 608
    },
    {
      "epoch": 1.7058823529411766,
      "grad_norm": 0.8783779740333557,
      "learning_rate": 3.851079694469636e-05,
      "loss": 0.2709,
      "step": 609
    },
    {
      "epoch": 1.708683473389356,
      "grad_norm": 0.23939639329910278,
      "learning_rate": 3.8472465897889394e-05,
      "loss": 0.1278,
      "step": 610
    },
    {
      "epoch": 1.7114845938375352,
      "grad_norm": 1.0147463083267212,
      "learning_rate": 3.843409016538716e-05,
      "loss": 0.2919,
      "step": 611
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.7156258225440979,
      "learning_rate": 3.8395669874474915e-05,
      "loss": 0.2452,
      "step": 612
    },
    {
      "epoch": 1.7170868347338937,
      "grad_norm": 0.23990198969841003,
      "learning_rate": 3.835720515258572e-05,
      "loss": 0.1744,
      "step": 613
    },
    {
      "epoch": 1.719887955182073,
      "grad_norm": 0.6649200320243835,
      "learning_rate": 3.831869612729999e-05,
      "loss": 0.2606,
      "step": 614
    },
    {
      "epoch": 1.7226890756302522,
      "grad_norm": 0.2535988390445709,
      "learning_rate": 3.828014292634509e-05,
      "loss": 0.1681,
      "step": 615
    },
    {
      "epoch": 1.7254901960784315,
      "grad_norm": 0.620143711566925,
      "learning_rate": 3.8241545677594895e-05,
      "loss": 0.2662,
      "step": 616
    },
    {
      "epoch": 1.7282913165266107,
      "grad_norm": 0.2728918492794037,
      "learning_rate": 3.820290450906941e-05,
      "loss": 0.1692,
      "step": 617
    },
    {
      "epoch": 1.73109243697479,
      "grad_norm": 0.2622087001800537,
      "learning_rate": 3.816421954893428e-05,
      "loss": 0.1215,
      "step": 618
    },
    {
      "epoch": 1.7338935574229692,
      "grad_norm": 0.3146950304508209,
      "learning_rate": 3.8125490925500425e-05,
      "loss": 0.1987,
      "step": 619
    },
    {
      "epoch": 1.7366946778711485,
      "grad_norm": 0.5234703421592712,
      "learning_rate": 3.808671876722357e-05,
      "loss": 0.2537,
      "step": 620
    },
    {
      "epoch": 1.7394957983193278,
      "grad_norm": 0.356790155172348,
      "learning_rate": 3.804790320270384e-05,
      "loss": 0.2111,
      "step": 621
    },
    {
      "epoch": 1.742296918767507,
      "grad_norm": 0.4688396155834198,
      "learning_rate": 3.800904436068533e-05,
      "loss": 0.0948,
      "step": 622
    },
    {
      "epoch": 1.7450980392156863,
      "grad_norm": 0.2965690493583679,
      "learning_rate": 3.797014237005571e-05,
      "loss": 0.1953,
      "step": 623
    },
    {
      "epoch": 1.7478991596638656,
      "grad_norm": 0.45940619707107544,
      "learning_rate": 3.793119735984572e-05,
      "loss": 0.2438,
      "step": 624
    },
    {
      "epoch": 1.7507002801120448,
      "grad_norm": 0.34540608525276184,
      "learning_rate": 3.78922094592288e-05,
      "loss": 0.1378,
      "step": 625
    },
    {
      "epoch": 1.753501400560224,
      "grad_norm": 0.2714684307575226,
      "learning_rate": 3.785317879752066e-05,
      "loss": 0.2209,
      "step": 626
    },
    {
      "epoch": 1.7563025210084033,
      "grad_norm": 0.6053199172019958,
      "learning_rate": 3.781410550417885e-05,
      "loss": 0.2887,
      "step": 627
    },
    {
      "epoch": 1.7591036414565826,
      "grad_norm": 0.3671252429485321,
      "learning_rate": 3.77749897088023e-05,
      "loss": 0.1822,
      "step": 628
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 0.2441740483045578,
      "learning_rate": 3.773583154113092e-05,
      "loss": 0.1712,
      "step": 629
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.5259542465209961,
      "learning_rate": 3.769663113104516e-05,
      "loss": 0.1065,
      "step": 630
    },
    {
      "epoch": 1.7675070028011204,
      "grad_norm": 0.2503260374069214,
      "learning_rate": 3.765738860856557e-05,
      "loss": 0.1761,
      "step": 631
    },
    {
      "epoch": 1.7703081232492996,
      "grad_norm": 0.5481187701225281,
      "learning_rate": 3.7618104103852415e-05,
      "loss": 0.2614,
      "step": 632
    },
    {
      "epoch": 1.773109243697479,
      "grad_norm": 0.27601489424705505,
      "learning_rate": 3.757877774720517e-05,
      "loss": 0.1901,
      "step": 633
    },
    {
      "epoch": 1.7759103641456582,
      "grad_norm": 0.2434099018573761,
      "learning_rate": 3.7539409669062136e-05,
      "loss": 0.205,
      "step": 634
    },
    {
      "epoch": 1.7787114845938374,
      "grad_norm": 0.28954920172691345,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.1636,
      "step": 635
    },
    {
      "epoch": 1.7815126050420167,
      "grad_norm": 0.5615999698638916,
      "learning_rate": 3.74605488707334e-05,
      "loss": 0.1061,
      "step": 636
    },
    {
      "epoch": 1.784313725490196,
      "grad_norm": 0.5544323325157166,
      "learning_rate": 3.742105641211449e-05,
      "loss": 0.106,
      "step": 637
    },
    {
      "epoch": 1.7871148459383752,
      "grad_norm": 0.5592769384384155,
      "learning_rate": 3.738152275513249e-05,
      "loss": 0.1079,
      "step": 638
    },
    {
      "epoch": 1.7899159663865545,
      "grad_norm": 0.2856994569301605,
      "learning_rate": 3.7341948030913294e-05,
      "loss": 0.1731,
      "step": 639
    },
    {
      "epoch": 1.7927170868347337,
      "grad_norm": 0.3745407164096832,
      "learning_rate": 3.730233237071898e-05,
      "loss": 0.1404,
      "step": 640
    },
    {
      "epoch": 1.795518207282913,
      "grad_norm": 0.33862021565437317,
      "learning_rate": 3.726267590594744e-05,
      "loss": 0.1952,
      "step": 641
    },
    {
      "epoch": 1.7983193277310925,
      "grad_norm": 0.3554513156414032,
      "learning_rate": 3.7222978768131854e-05,
      "loss": 0.1322,
      "step": 642
    },
    {
      "epoch": 1.8011204481792717,
      "grad_norm": 0.22815661132335663,
      "learning_rate": 3.718324108894036e-05,
      "loss": 0.1714,
      "step": 643
    },
    {
      "epoch": 1.803921568627451,
      "grad_norm": 0.32825687527656555,
      "learning_rate": 3.7143463000175546e-05,
      "loss": 0.1421,
      "step": 644
    },
    {
      "epoch": 1.8067226890756303,
      "grad_norm": 0.24787546694278717,
      "learning_rate": 3.7103644633774014e-05,
      "loss": 0.1682,
      "step": 645
    },
    {
      "epoch": 1.8095238095238095,
      "grad_norm": 0.3160790801048279,
      "learning_rate": 3.706378612180598e-05,
      "loss": 0.193,
      "step": 646
    },
    {
      "epoch": 1.8123249299719888,
      "grad_norm": 0.22702836990356445,
      "learning_rate": 3.70238875964748e-05,
      "loss": 0.171,
      "step": 647
    },
    {
      "epoch": 1.815126050420168,
      "grad_norm": 0.2879033386707306,
      "learning_rate": 3.6983949190116576e-05,
      "loss": 0.1391,
      "step": 648
    },
    {
      "epoch": 1.8179271708683473,
      "grad_norm": 0.2816312313079834,
      "learning_rate": 3.6943971035199645e-05,
      "loss": 0.1351,
      "step": 649
    },
    {
      "epoch": 1.8207282913165266,
      "grad_norm": 0.2494056075811386,
      "learning_rate": 3.690395326432421e-05,
      "loss": 0.165,
      "step": 650
    },
    {
      "epoch": 1.8235294117647058,
      "grad_norm": 0.5620803236961365,
      "learning_rate": 3.686389601022188e-05,
      "loss": 0.256,
      "step": 651
    },
    {
      "epoch": 1.826330532212885,
      "grad_norm": 0.36324793100357056,
      "learning_rate": 3.682379940575519e-05,
      "loss": 0.1187,
      "step": 652
    },
    {
      "epoch": 1.8291316526610646,
      "grad_norm": 0.32306012511253357,
      "learning_rate": 3.678366358391723e-05,
      "loss": 0.1183,
      "step": 653
    },
    {
      "epoch": 1.8319327731092439,
      "grad_norm": 0.25600212812423706,
      "learning_rate": 3.674348867783115e-05,
      "loss": 0.1684,
      "step": 654
    },
    {
      "epoch": 1.8347338935574231,
      "grad_norm": 0.27342215180397034,
      "learning_rate": 3.670327482074973e-05,
      "loss": 0.1249,
      "step": 655
    },
    {
      "epoch": 1.8375350140056024,
      "grad_norm": 0.4090156555175781,
      "learning_rate": 3.666302214605495e-05,
      "loss": 0.179,
      "step": 656
    },
    {
      "epoch": 1.8403361344537816,
      "grad_norm": 0.4678898751735687,
      "learning_rate": 3.662273078725754e-05,
      "loss": 0.2021,
      "step": 657
    },
    {
      "epoch": 1.843137254901961,
      "grad_norm": 0.28741687536239624,
      "learning_rate": 3.6582400877996546e-05,
      "loss": 0.1641,
      "step": 658
    },
    {
      "epoch": 1.8459383753501402,
      "grad_norm": 0.4612816572189331,
      "learning_rate": 3.654203255203886e-05,
      "loss": 0.2167,
      "step": 659
    },
    {
      "epoch": 1.8487394957983194,
      "grad_norm": 0.27487486600875854,
      "learning_rate": 3.6501625943278805e-05,
      "loss": 0.1567,
      "step": 660
    },
    {
      "epoch": 1.8515406162464987,
      "grad_norm": 0.4421831965446472,
      "learning_rate": 3.6461181185737694e-05,
      "loss": 0.2094,
      "step": 661
    },
    {
      "epoch": 1.854341736694678,
      "grad_norm": 0.5941998362541199,
      "learning_rate": 3.6420698413563345e-05,
      "loss": 0.2084,
      "step": 662
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.5112701058387756,
      "learning_rate": 3.6380177761029685e-05,
      "loss": 0.0836,
      "step": 663
    },
    {
      "epoch": 1.8599439775910365,
      "grad_norm": 0.8637323975563049,
      "learning_rate": 3.633961936253628e-05,
      "loss": 0.2437,
      "step": 664
    },
    {
      "epoch": 1.8627450980392157,
      "grad_norm": 0.4770047962665558,
      "learning_rate": 3.629902335260789e-05,
      "loss": 0.1965,
      "step": 665
    },
    {
      "epoch": 1.865546218487395,
      "grad_norm": 0.3599564731121063,
      "learning_rate": 3.625838986589403e-05,
      "loss": 0.1134,
      "step": 666
    },
    {
      "epoch": 1.8683473389355743,
      "grad_norm": 0.5263693928718567,
      "learning_rate": 3.621771903716849e-05,
      "loss": 0.0875,
      "step": 667
    },
    {
      "epoch": 1.8711484593837535,
      "grad_norm": 0.5898877382278442,
      "learning_rate": 3.617701100132897e-05,
      "loss": 0.2261,
      "step": 668
    },
    {
      "epoch": 1.8739495798319328,
      "grad_norm": 0.369739830493927,
      "learning_rate": 3.613626589339653e-05,
      "loss": 0.1718,
      "step": 669
    },
    {
      "epoch": 1.876750700280112,
      "grad_norm": 0.368196576833725,
      "learning_rate": 3.609548384851522e-05,
      "loss": 0.194,
      "step": 670
    },
    {
      "epoch": 1.8795518207282913,
      "grad_norm": 0.2884950339794159,
      "learning_rate": 3.605466500195159e-05,
      "loss": 0.1578,
      "step": 671
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.7132876515388489,
      "learning_rate": 3.601380948909425e-05,
      "loss": 0.2391,
      "step": 672
    },
    {
      "epoch": 1.8851540616246498,
      "grad_norm": 0.29729971289634705,
      "learning_rate": 3.597291744545344e-05,
      "loss": 0.1518,
      "step": 673
    },
    {
      "epoch": 1.887955182072829,
      "grad_norm": 0.24395136535167694,
      "learning_rate": 3.593198900666056e-05,
      "loss": 0.1675,
      "step": 674
    },
    {
      "epoch": 1.8907563025210083,
      "grad_norm": 0.28539448976516724,
      "learning_rate": 3.589102430846773e-05,
      "loss": 0.1573,
      "step": 675
    },
    {
      "epoch": 1.8935574229691876,
      "grad_norm": 0.38136017322540283,
      "learning_rate": 3.585002348674733e-05,
      "loss": 0.2004,
      "step": 676
    },
    {
      "epoch": 1.8963585434173669,
      "grad_norm": 0.5776926279067993,
      "learning_rate": 3.5808986677491555e-05,
      "loss": 0.0913,
      "step": 677
    },
    {
      "epoch": 1.8991596638655461,
      "grad_norm": 0.4445839524269104,
      "learning_rate": 3.576791401681194e-05,
      "loss": 0.185,
      "step": 678
    },
    {
      "epoch": 1.9019607843137254,
      "grad_norm": 0.31285786628723145,
      "learning_rate": 3.5726805640939e-05,
      "loss": 0.1609,
      "step": 679
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 0.38080793619155884,
      "learning_rate": 3.5685661686221644e-05,
      "loss": 0.1928,
      "step": 680
    },
    {
      "epoch": 1.907563025210084,
      "grad_norm": 0.37026283144950867,
      "learning_rate": 3.564448228912682e-05,
      "loss": 0.1829,
      "step": 681
    },
    {
      "epoch": 1.9103641456582632,
      "grad_norm": 0.3265126347541809,
      "learning_rate": 3.5603267586239026e-05,
      "loss": 0.1563,
      "step": 682
    },
    {
      "epoch": 1.9131652661064424,
      "grad_norm": 0.6355715990066528,
      "learning_rate": 3.556201771425985e-05,
      "loss": 0.2228,
      "step": 683
    },
    {
      "epoch": 1.9159663865546217,
      "grad_norm": 0.5335772037506104,
      "learning_rate": 3.5520732810007566e-05,
      "loss": 0.1654,
      "step": 684
    },
    {
      "epoch": 1.918767507002801,
      "grad_norm": 0.6501393914222717,
      "learning_rate": 3.547941301041661e-05,
      "loss": 0.097,
      "step": 685
    },
    {
      "epoch": 1.9215686274509802,
      "grad_norm": 0.6850645542144775,
      "learning_rate": 3.543805845253716e-05,
      "loss": 0.1012,
      "step": 686
    },
    {
      "epoch": 1.9243697478991597,
      "grad_norm": 0.659045398235321,
      "learning_rate": 3.539666927353469e-05,
      "loss": 0.0977,
      "step": 687
    },
    {
      "epoch": 1.927170868347339,
      "grad_norm": 0.47713983058929443,
      "learning_rate": 3.535524561068952e-05,
      "loss": 0.1192,
      "step": 688
    },
    {
      "epoch": 1.9299719887955182,
      "grad_norm": 0.4114561975002289,
      "learning_rate": 3.5313787601396325e-05,
      "loss": 0.1243,
      "step": 689
    },
    {
      "epoch": 1.9327731092436975,
      "grad_norm": 0.4006536900997162,
      "learning_rate": 3.527229538316371e-05,
      "loss": 0.1811,
      "step": 690
    },
    {
      "epoch": 1.9355742296918768,
      "grad_norm": 0.520041823387146,
      "learning_rate": 3.523076909361373e-05,
      "loss": 0.2083,
      "step": 691
    },
    {
      "epoch": 1.938375350140056,
      "grad_norm": 0.3322596251964569,
      "learning_rate": 3.518920887048149e-05,
      "loss": 0.1568,
      "step": 692
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.3033127188682556,
      "learning_rate": 3.514761485161458e-05,
      "loss": 0.1544,
      "step": 693
    },
    {
      "epoch": 1.9439775910364145,
      "grad_norm": 1.5677649974822998,
      "learning_rate": 3.510598717497276e-05,
      "loss": 0.3283,
      "step": 694
    },
    {
      "epoch": 1.9467787114845938,
      "grad_norm": 0.9142319560050964,
      "learning_rate": 3.506432597862737e-05,
      "loss": 0.2276,
      "step": 695
    },
    {
      "epoch": 1.949579831932773,
      "grad_norm": 0.3185567259788513,
      "learning_rate": 3.5022631400760944e-05,
      "loss": 0.1542,
      "step": 696
    },
    {
      "epoch": 1.9523809523809523,
      "grad_norm": 0.6337863802909851,
      "learning_rate": 3.4980903579666744e-05,
      "loss": 0.2202,
      "step": 697
    },
    {
      "epoch": 1.9551820728291318,
      "grad_norm": 0.3815496861934662,
      "learning_rate": 3.493914265374829e-05,
      "loss": 0.1303,
      "step": 698
    },
    {
      "epoch": 1.957983193277311,
      "grad_norm": 0.4644532799720764,
      "learning_rate": 3.489734876151891e-05,
      "loss": 0.1881,
      "step": 699
    },
    {
      "epoch": 1.9607843137254903,
      "grad_norm": 0.43058714270591736,
      "learning_rate": 3.485552204160126e-05,
      "loss": 0.1121,
      "step": 700
    },
    {
      "epoch": 1.9635854341736696,
      "grad_norm": 1.0429691076278687,
      "learning_rate": 3.48136626327269e-05,
      "loss": 0.2764,
      "step": 701
    },
    {
      "epoch": 1.9663865546218489,
      "grad_norm": 0.6157497763633728,
      "learning_rate": 3.4771770673735796e-05,
      "loss": 0.0843,
      "step": 702
    },
    {
      "epoch": 1.9691876750700281,
      "grad_norm": 0.33218902349472046,
      "learning_rate": 3.472984630357587e-05,
      "loss": 0.1626,
      "step": 703
    },
    {
      "epoch": 1.9719887955182074,
      "grad_norm": 0.6079518795013428,
      "learning_rate": 3.4687889661302576e-05,
      "loss": 0.0859,
      "step": 704
    },
    {
      "epoch": 1.9747899159663866,
      "grad_norm": 0.2988344728946686,
      "learning_rate": 3.464590088607839e-05,
      "loss": 0.1604,
      "step": 705
    },
    {
      "epoch": 1.977591036414566,
      "grad_norm": 0.47207456827163696,
      "learning_rate": 3.460388011717236e-05,
      "loss": 0.1837,
      "step": 706
    },
    {
      "epoch": 1.9803921568627452,
      "grad_norm": 0.35358136892318726,
      "learning_rate": 3.456182749395966e-05,
      "loss": 0.1482,
      "step": 707
    },
    {
      "epoch": 1.9831932773109244,
      "grad_norm": 0.405224084854126,
      "learning_rate": 3.451974315592113e-05,
      "loss": 0.1288,
      "step": 708
    },
    {
      "epoch": 1.9859943977591037,
      "grad_norm": 0.4823918640613556,
      "learning_rate": 3.4477627242642776e-05,
      "loss": 0.1835,
      "step": 709
    },
    {
      "epoch": 1.988795518207283,
      "grad_norm": 0.3758796155452728,
      "learning_rate": 3.443547989381536e-05,
      "loss": 0.1303,
      "step": 710
    },
    {
      "epoch": 1.9915966386554622,
      "grad_norm": 0.47359490394592285,
      "learning_rate": 3.43933012492339e-05,
      "loss": 0.1844,
      "step": 711
    },
    {
      "epoch": 1.9943977591036415,
      "grad_norm": 0.3807266652584076,
      "learning_rate": 3.43510914487972e-05,
      "loss": 0.1357,
      "step": 712
    },
    {
      "epoch": 1.9971988795518207,
      "grad_norm": 0.6917422413825989,
      "learning_rate": 3.430885063250743e-05,
      "loss": 0.2231,
      "step": 713
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6366453766822815,
      "learning_rate": 3.42665789404696e-05,
      "loss": 0.0794,
      "step": 714
    },
    {
      "epoch": 2.0,
      "eval_f1 (minor class)": 0.11274509803921569,
      "eval_loss": 0.17888100445270538,
      "eval_roc_auc": 0.5086586164934943,
      "eval_runtime": 2.9373,
      "eval_samples_per_second": 431.683,
      "eval_steps_per_second": 13.618,
      "step": 714
    },
    {
      "epoch": 2.0028011204481793,
      "grad_norm": 0.6449307203292847,
      "learning_rate": 3.422427651289118e-05,
      "loss": 0.0814,
      "step": 715
    },
    {
      "epoch": 2.0056022408963585,
      "grad_norm": 1.047551155090332,
      "learning_rate": 3.418194349008153e-05,
      "loss": 0.2542,
      "step": 716
    },
    {
      "epoch": 2.008403361344538,
      "grad_norm": 0.486785888671875,
      "learning_rate": 3.413958001245152e-05,
      "loss": 0.1403,
      "step": 717
    },
    {
      "epoch": 2.011204481792717,
      "grad_norm": 1.1322331428527832,
      "learning_rate": 3.409718622051303e-05,
      "loss": 0.2194,
      "step": 718
    },
    {
      "epoch": 2.0140056022408963,
      "grad_norm": 0.41819918155670166,
      "learning_rate": 3.4054762254878476e-05,
      "loss": 0.151,
      "step": 719
    },
    {
      "epoch": 2.0168067226890756,
      "grad_norm": 0.38247114419937134,
      "learning_rate": 3.401230825626037e-05,
      "loss": 0.1593,
      "step": 720
    },
    {
      "epoch": 2.019607843137255,
      "grad_norm": 0.7601361274719238,
      "learning_rate": 3.396982436547082e-05,
      "loss": 0.2249,
      "step": 721
    },
    {
      "epoch": 2.022408963585434,
      "grad_norm": 1.248322606086731,
      "learning_rate": 3.392731072342109e-05,
      "loss": 0.197,
      "step": 722
    },
    {
      "epoch": 2.0252100840336134,
      "grad_norm": 0.5805521607398987,
      "learning_rate": 3.388476747112113e-05,
      "loss": 0.1443,
      "step": 723
    },
    {
      "epoch": 2.0280112044817926,
      "grad_norm": 0.47655874490737915,
      "learning_rate": 3.384219474967908e-05,
      "loss": 0.1962,
      "step": 724
    },
    {
      "epoch": 2.030812324929972,
      "grad_norm": 0.48123157024383545,
      "learning_rate": 3.3799592700300866e-05,
      "loss": 0.1266,
      "step": 725
    },
    {
      "epoch": 2.033613445378151,
      "grad_norm": 0.9366979002952576,
      "learning_rate": 3.375696146428963e-05,
      "loss": 0.2132,
      "step": 726
    },
    {
      "epoch": 2.0364145658263304,
      "grad_norm": 0.38628774881362915,
      "learning_rate": 3.3714301183045385e-05,
      "loss": 0.1599,
      "step": 727
    },
    {
      "epoch": 2.0392156862745097,
      "grad_norm": 0.5700298547744751,
      "learning_rate": 3.3671611998064425e-05,
      "loss": 0.1271,
      "step": 728
    },
    {
      "epoch": 2.042016806722689,
      "grad_norm": 0.8660473823547363,
      "learning_rate": 3.3628894050938945e-05,
      "loss": 0.2301,
      "step": 729
    },
    {
      "epoch": 2.044817927170868,
      "grad_norm": 0.7136172652244568,
      "learning_rate": 3.3586147483356535e-05,
      "loss": 0.1177,
      "step": 730
    },
    {
      "epoch": 2.0476190476190474,
      "grad_norm": 0.727861225605011,
      "learning_rate": 3.354337243709971e-05,
      "loss": 0.2178,
      "step": 731
    },
    {
      "epoch": 2.0504201680672267,
      "grad_norm": 1.5354722738265991,
      "learning_rate": 3.350056905404543e-05,
      "loss": 0.3138,
      "step": 732
    },
    {
      "epoch": 2.053221288515406,
      "grad_norm": 0.6516889333724976,
      "learning_rate": 3.345773747616467e-05,
      "loss": 0.221,
      "step": 733
    },
    {
      "epoch": 2.0560224089635852,
      "grad_norm": 0.4907413125038147,
      "learning_rate": 3.34148778455219e-05,
      "loss": 0.1524,
      "step": 734
    },
    {
      "epoch": 2.0588235294117645,
      "grad_norm": 0.5165950059890747,
      "learning_rate": 3.3371990304274656e-05,
      "loss": 0.199,
      "step": 735
    },
    {
      "epoch": 2.0616246498599438,
      "grad_norm": 1.439732313156128,
      "learning_rate": 3.332907499467302e-05,
      "loss": 0.2949,
      "step": 736
    },
    {
      "epoch": 2.064425770308123,
      "grad_norm": 0.650676965713501,
      "learning_rate": 3.328613205905921e-05,
      "loss": 0.145,
      "step": 737
    },
    {
      "epoch": 2.0672268907563027,
      "grad_norm": 0.444158673286438,
      "learning_rate": 3.324316163986704e-05,
      "loss": 0.219,
      "step": 738
    },
    {
      "epoch": 2.070028011204482,
      "grad_norm": 0.7450339794158936,
      "learning_rate": 3.320016387962151e-05,
      "loss": 0.1589,
      "step": 739
    },
    {
      "epoch": 2.0728291316526612,
      "grad_norm": 1.0164347887039185,
      "learning_rate": 3.315713892093829e-05,
      "loss": 0.1207,
      "step": 740
    },
    {
      "epoch": 2.0756302521008405,
      "grad_norm": 0.6916231513023376,
      "learning_rate": 3.3114086906523266e-05,
      "loss": 0.17,
      "step": 741
    },
    {
      "epoch": 2.0784313725490198,
      "grad_norm": 1.029963493347168,
      "learning_rate": 3.307100797917207e-05,
      "loss": 0.1218,
      "step": 742
    },
    {
      "epoch": 2.081232492997199,
      "grad_norm": 0.8726139068603516,
      "learning_rate": 3.302790228176959e-05,
      "loss": 0.1337,
      "step": 743
    },
    {
      "epoch": 2.0840336134453783,
      "grad_norm": 0.6706817746162415,
      "learning_rate": 3.29847699572895e-05,
      "loss": 0.1637,
      "step": 744
    },
    {
      "epoch": 2.0868347338935576,
      "grad_norm": 0.7627484798431396,
      "learning_rate": 3.294161114879382e-05,
      "loss": 0.2022,
      "step": 745
    },
    {
      "epoch": 2.089635854341737,
      "grad_norm": 0.6804760098457336,
      "learning_rate": 3.289842599943237e-05,
      "loss": 0.1245,
      "step": 746
    },
    {
      "epoch": 2.092436974789916,
      "grad_norm": 0.5034189224243164,
      "learning_rate": 3.285521465244237e-05,
      "loss": 0.141,
      "step": 747
    },
    {
      "epoch": 2.0952380952380953,
      "grad_norm": 0.9262709617614746,
      "learning_rate": 3.281197725114792e-05,
      "loss": 0.099,
      "step": 748
    },
    {
      "epoch": 2.0980392156862746,
      "grad_norm": 0.45192310214042664,
      "learning_rate": 3.276871393895954e-05,
      "loss": 0.1634,
      "step": 749
    },
    {
      "epoch": 2.100840336134454,
      "grad_norm": 1.2695709466934204,
      "learning_rate": 3.272542485937369e-05,
      "loss": 0.2591,
      "step": 750
    },
    {
      "epoch": 2.103641456582633,
      "grad_norm": 0.6426392197608948,
      "learning_rate": 3.26821101559723e-05,
      "loss": 0.1701,
      "step": 751
    },
    {
      "epoch": 2.1064425770308124,
      "grad_norm": 0.44048747420310974,
      "learning_rate": 3.263876997242229e-05,
      "loss": 0.1532,
      "step": 752
    },
    {
      "epoch": 2.1092436974789917,
      "grad_norm": 1.7888472080230713,
      "learning_rate": 3.259540445247509e-05,
      "loss": 0.2987,
      "step": 753
    },
    {
      "epoch": 2.112044817927171,
      "grad_norm": 1.113670825958252,
      "learning_rate": 3.2552013739966145e-05,
      "loss": 0.2087,
      "step": 754
    },
    {
      "epoch": 2.11484593837535,
      "grad_norm": 0.6354279518127441,
      "learning_rate": 3.250859797881452e-05,
      "loss": 0.1307,
      "step": 755
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 0.5894302129745483,
      "learning_rate": 3.2465157313022284e-05,
      "loss": 0.094,
      "step": 756
    },
    {
      "epoch": 2.1204481792717087,
      "grad_norm": 0.5198976397514343,
      "learning_rate": 3.242169188667416e-05,
      "loss": 0.1224,
      "step": 757
    },
    {
      "epoch": 2.123249299719888,
      "grad_norm": 0.4988885521888733,
      "learning_rate": 3.2378201843937e-05,
      "loss": 0.1241,
      "step": 758
    },
    {
      "epoch": 2.1260504201680672,
      "grad_norm": 1.0682430267333984,
      "learning_rate": 3.233468732905927e-05,
      "loss": 0.2033,
      "step": 759
    },
    {
      "epoch": 2.1288515406162465,
      "grad_norm": 1.5533567667007446,
      "learning_rate": 3.2291148486370626e-05,
      "loss": 0.2584,
      "step": 760
    },
    {
      "epoch": 2.1316526610644257,
      "grad_norm": 0.7770599126815796,
      "learning_rate": 3.224758546028143e-05,
      "loss": 0.0703,
      "step": 761
    },
    {
      "epoch": 2.134453781512605,
      "grad_norm": 0.8279226422309875,
      "learning_rate": 3.220399839528222e-05,
      "loss": 0.0776,
      "step": 762
    },
    {
      "epoch": 2.1372549019607843,
      "grad_norm": 1.2002817392349243,
      "learning_rate": 3.2160387435943296e-05,
      "loss": 0.2055,
      "step": 763
    },
    {
      "epoch": 2.1400560224089635,
      "grad_norm": 0.5353344678878784,
      "learning_rate": 3.21167527269142e-05,
      "loss": 0.1137,
      "step": 764
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 0.7878655791282654,
      "learning_rate": 3.207309441292325e-05,
      "loss": 0.0716,
      "step": 765
    },
    {
      "epoch": 2.145658263305322,
      "grad_norm": 0.562883198261261,
      "learning_rate": 3.202941263877706e-05,
      "loss": 0.0879,
      "step": 766
    },
    {
      "epoch": 2.1484593837535013,
      "grad_norm": 1.004749059677124,
      "learning_rate": 3.198570754936004e-05,
      "loss": 0.1883,
      "step": 767
    },
    {
      "epoch": 2.1512605042016806,
      "grad_norm": 0.8928613066673279,
      "learning_rate": 3.194197928963396e-05,
      "loss": 0.164,
      "step": 768
    },
    {
      "epoch": 2.15406162464986,
      "grad_norm": 2.246786594390869,
      "learning_rate": 3.189822800463742e-05,
      "loss": 0.1953,
      "step": 769
    },
    {
      "epoch": 2.156862745098039,
      "grad_norm": 1.1755541563034058,
      "learning_rate": 3.1854453839485385e-05,
      "loss": 0.1598,
      "step": 770
    },
    {
      "epoch": 2.1596638655462184,
      "grad_norm": 0.8687044382095337,
      "learning_rate": 3.1810656939368744e-05,
      "loss": 0.0719,
      "step": 771
    },
    {
      "epoch": 2.1624649859943976,
      "grad_norm": 1.5823578834533691,
      "learning_rate": 3.176683744955377e-05,
      "loss": 0.1969,
      "step": 772
    },
    {
      "epoch": 2.165266106442577,
      "grad_norm": 0.6787047982215881,
      "learning_rate": 3.172299551538164e-05,
      "loss": 0.1161,
      "step": 773
    },
    {
      "epoch": 2.168067226890756,
      "grad_norm": 2.3936171531677246,
      "learning_rate": 3.167913128226803e-05,
      "loss": 0.2243,
      "step": 774
    },
    {
      "epoch": 2.1708683473389354,
      "grad_norm": 0.9804347157478333,
      "learning_rate": 3.1635244895702524e-05,
      "loss": 0.0667,
      "step": 775
    },
    {
      "epoch": 2.1736694677871147,
      "grad_norm": 1.6783931255340576,
      "learning_rate": 3.159133650124821e-05,
      "loss": 0.177,
      "step": 776
    },
    {
      "epoch": 2.176470588235294,
      "grad_norm": 1.7065593004226685,
      "learning_rate": 3.1547406244541175e-05,
      "loss": 0.2228,
      "step": 777
    },
    {
      "epoch": 2.179271708683473,
      "grad_norm": 1.4685626029968262,
      "learning_rate": 3.150345427129002e-05,
      "loss": 0.212,
      "step": 778
    },
    {
      "epoch": 2.1820728291316525,
      "grad_norm": 1.0432506799697876,
      "learning_rate": 3.145948072727535e-05,
      "loss": 0.1374,
      "step": 779
    },
    {
      "epoch": 2.184873949579832,
      "grad_norm": 1.9101930856704712,
      "learning_rate": 3.1415485758349346e-05,
      "loss": 0.2451,
      "step": 780
    },
    {
      "epoch": 2.1876750700280114,
      "grad_norm": 1.068935513496399,
      "learning_rate": 3.137146951043524e-05,
      "loss": 0.1353,
      "step": 781
    },
    {
      "epoch": 2.1904761904761907,
      "grad_norm": 0.9940873980522156,
      "learning_rate": 3.132743212952684e-05,
      "loss": 0.0726,
      "step": 782
    },
    {
      "epoch": 2.19327731092437,
      "grad_norm": 1.092490315437317,
      "learning_rate": 3.128337376168805e-05,
      "loss": 0.174,
      "step": 783
    },
    {
      "epoch": 2.196078431372549,
      "grad_norm": 0.75323885679245,
      "learning_rate": 3.123929455305239e-05,
      "loss": 0.1166,
      "step": 784
    },
    {
      "epoch": 2.1988795518207285,
      "grad_norm": 1.346554160118103,
      "learning_rate": 3.11951946498225e-05,
      "loss": 0.1646,
      "step": 785
    },
    {
      "epoch": 2.2016806722689077,
      "grad_norm": 0.9711296558380127,
      "learning_rate": 3.115107419826966e-05,
      "loss": 0.1301,
      "step": 786
    },
    {
      "epoch": 2.204481792717087,
      "grad_norm": 0.9502242207527161,
      "learning_rate": 3.1106933344733304e-05,
      "loss": 0.1036,
      "step": 787
    },
    {
      "epoch": 2.2072829131652663,
      "grad_norm": 1.8469451665878296,
      "learning_rate": 3.106277223562054e-05,
      "loss": 0.2247,
      "step": 788
    },
    {
      "epoch": 2.2100840336134455,
      "grad_norm": 0.8027406334877014,
      "learning_rate": 3.101859101740564e-05,
      "loss": 0.1404,
      "step": 789
    },
    {
      "epoch": 2.212885154061625,
      "grad_norm": 1.0038503408432007,
      "learning_rate": 3.0974389836629624e-05,
      "loss": 0.1889,
      "step": 790
    },
    {
      "epoch": 2.215686274509804,
      "grad_norm": 2.352829933166504,
      "learning_rate": 3.0930168839899666e-05,
      "loss": 0.2958,
      "step": 791
    },
    {
      "epoch": 2.2184873949579833,
      "grad_norm": 0.6628269553184509,
      "learning_rate": 3.088592817388869e-05,
      "loss": 0.1324,
      "step": 792
    },
    {
      "epoch": 2.2212885154061626,
      "grad_norm": 0.7565025091171265,
      "learning_rate": 3.084166798533489e-05,
      "loss": 0.1405,
      "step": 793
    },
    {
      "epoch": 2.224089635854342,
      "grad_norm": 0.8182944655418396,
      "learning_rate": 3.079738842104115e-05,
      "loss": 0.1145,
      "step": 794
    },
    {
      "epoch": 2.226890756302521,
      "grad_norm": 0.6144469380378723,
      "learning_rate": 3.075308962787466e-05,
      "loss": 0.1457,
      "step": 795
    },
    {
      "epoch": 2.2296918767507004,
      "grad_norm": 1.0344878435134888,
      "learning_rate": 3.0708771752766394e-05,
      "loss": 0.118,
      "step": 796
    },
    {
      "epoch": 2.2324929971988796,
      "grad_norm": 0.6334454417228699,
      "learning_rate": 3.06644349427106e-05,
      "loss": 0.137,
      "step": 797
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 1.0241148471832275,
      "learning_rate": 3.062007934476433e-05,
      "loss": 0.073,
      "step": 798
    },
    {
      "epoch": 2.238095238095238,
      "grad_norm": 1.3902757167816162,
      "learning_rate": 3.057570510604696e-05,
      "loss": 0.1635,
      "step": 799
    },
    {
      "epoch": 2.2408963585434174,
      "grad_norm": 1.1507909297943115,
      "learning_rate": 3.05313123737397e-05,
      "loss": 0.1311,
      "step": 800
    }
  ],
  "logging_steps": 1,
  "max_steps": 1785,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 192309268647936.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}