{
  "best_metric": 1.8645277672129885,
  "best_model_checkpoint": "./modernBERT-content-regression/run-3/checkpoint-248",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 248,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008064516129032258,
      "grad_norm": 312.8093566894531,
      "learning_rate": 1.1731316602713056e-06,
      "loss": 21.3087,
      "step": 1
    },
    {
      "epoch": 0.016129032258064516,
      "grad_norm": 97.11482238769531,
      "learning_rate": 1.1712364556505119e-06,
      "loss": 1.9024,
      "step": 2
    },
    {
      "epoch": 0.024193548387096774,
      "grad_norm": 608.1284790039062,
      "learning_rate": 1.1693412510297182e-06,
      "loss": 154.2055,
      "step": 3
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 205.3589630126953,
      "learning_rate": 1.1674460464089244e-06,
      "loss": 18.409,
      "step": 4
    },
    {
      "epoch": 0.04032258064516129,
      "grad_norm": 116.6544189453125,
      "learning_rate": 1.165550841788131e-06,
      "loss": 4.2643,
      "step": 5
    },
    {
      "epoch": 0.04838709677419355,
      "grad_norm": 34.426883697509766,
      "learning_rate": 1.1636556371673372e-06,
      "loss": 0.5243,
      "step": 6
    },
    {
      "epoch": 0.056451612903225805,
      "grad_norm": 161.2743682861328,
      "learning_rate": 1.1617604325465434e-06,
      "loss": 9.653,
      "step": 7
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 316.244384765625,
      "learning_rate": 1.1598652279257497e-06,
      "loss": 21.3918,
      "step": 8
    },
    {
      "epoch": 0.07258064516129033,
      "grad_norm": 64.28992462158203,
      "learning_rate": 1.157970023304956e-06,
      "loss": 0.6204,
      "step": 9
    },
    {
      "epoch": 0.08064516129032258,
      "grad_norm": 130.6162109375,
      "learning_rate": 1.1560748186841622e-06,
      "loss": 5.0316,
      "step": 10
    },
    {
      "epoch": 0.08870967741935484,
      "grad_norm": 340.1315612792969,
      "learning_rate": 1.1541796140633685e-06,
      "loss": 18.7971,
      "step": 11
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 63.929317474365234,
      "learning_rate": 1.1522844094425748e-06,
      "loss": 1.2454,
      "step": 12
    },
    {
      "epoch": 0.10483870967741936,
      "grad_norm": 5.4183030128479,
      "learning_rate": 1.150389204821781e-06,
      "loss": 0.0884,
      "step": 13
    },
    {
      "epoch": 0.11290322580645161,
      "grad_norm": 137.7165985107422,
      "learning_rate": 1.1484940002009875e-06,
      "loss": 4.2256,
      "step": 14
    },
    {
      "epoch": 0.12096774193548387,
      "grad_norm": 31.921754837036133,
      "learning_rate": 1.1465987955801938e-06,
      "loss": 0.3319,
      "step": 15
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 200.37228393554688,
      "learning_rate": 1.1447035909594e-06,
      "loss": 11.9662,
      "step": 16
    },
    {
      "epoch": 0.13709677419354838,
      "grad_norm": 54.43822479248047,
      "learning_rate": 1.1428083863386063e-06,
      "loss": 1.2817,
      "step": 17
    },
    {
      "epoch": 0.14516129032258066,
      "grad_norm": 122.57898712158203,
      "learning_rate": 1.1409131817178126e-06,
      "loss": 3.3717,
      "step": 18
    },
    {
      "epoch": 0.1532258064516129,
      "grad_norm": 24.417898178100586,
      "learning_rate": 1.1390179770970188e-06,
      "loss": 0.5961,
      "step": 19
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 280.314453125,
      "learning_rate": 1.1371227724762251e-06,
      "loss": 12.7171,
      "step": 20
    },
    {
      "epoch": 0.1693548387096774,
      "grad_norm": 74.58489227294922,
      "learning_rate": 1.1352275678554316e-06,
      "loss": 0.6804,
      "step": 21
    },
    {
      "epoch": 0.1774193548387097,
      "grad_norm": 134.18431091308594,
      "learning_rate": 1.1333323632346376e-06,
      "loss": 1.4807,
      "step": 22
    },
    {
      "epoch": 0.18548387096774194,
      "grad_norm": 127.26359558105469,
      "learning_rate": 1.1314371586138441e-06,
      "loss": 18.1665,
      "step": 23
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 196.29579162597656,
      "learning_rate": 1.1295419539930504e-06,
      "loss": 2.6718,
      "step": 24
    },
    {
      "epoch": 0.20161290322580644,
      "grad_norm": 59.49917984008789,
      "learning_rate": 1.1276467493722567e-06,
      "loss": 0.7888,
      "step": 25
    },
    {
      "epoch": 0.20967741935483872,
      "grad_norm": 160.57215881347656,
      "learning_rate": 1.125751544751463e-06,
      "loss": 0.7413,
      "step": 26
    },
    {
      "epoch": 0.21774193548387097,
      "grad_norm": 63.3122444152832,
      "learning_rate": 1.1238563401306692e-06,
      "loss": 3.04,
      "step": 27
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 132.4581756591797,
      "learning_rate": 1.1219611355098755e-06,
      "loss": 1.7609,
      "step": 28
    },
    {
      "epoch": 0.23387096774193547,
      "grad_norm": 69.83406829833984,
      "learning_rate": 1.1200659308890817e-06,
      "loss": 1.1596,
      "step": 29
    },
    {
      "epoch": 0.24193548387096775,
      "grad_norm": 34.646114349365234,
      "learning_rate": 1.1181707262682882e-06,
      "loss": 0.5233,
      "step": 30
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.260757446289062,
      "learning_rate": 1.1162755216474943e-06,
      "loss": 1.0132,
      "step": 31
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 44.67536544799805,
      "learning_rate": 1.1143803170267007e-06,
      "loss": 0.9895,
      "step": 32
    },
    {
      "epoch": 0.2661290322580645,
      "grad_norm": 54.21400451660156,
      "learning_rate": 1.112485112405907e-06,
      "loss": 1.1101,
      "step": 33
    },
    {
      "epoch": 0.27419354838709675,
      "grad_norm": 36.84498977661133,
      "learning_rate": 1.1105899077851133e-06,
      "loss": 1.3959,
      "step": 34
    },
    {
      "epoch": 0.28225806451612906,
      "grad_norm": 369.2140808105469,
      "learning_rate": 1.1086947031643195e-06,
      "loss": 23.5792,
      "step": 35
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 51.993858337402344,
      "learning_rate": 1.1067994985435258e-06,
      "loss": 0.6532,
      "step": 36
    },
    {
      "epoch": 0.29838709677419356,
      "grad_norm": 87.89725494384766,
      "learning_rate": 1.104904293922732e-06,
      "loss": 1.0406,
      "step": 37
    },
    {
      "epoch": 0.3064516129032258,
      "grad_norm": 272.9493713378906,
      "learning_rate": 1.1030090893019383e-06,
      "loss": 24.8131,
      "step": 38
    },
    {
      "epoch": 0.31451612903225806,
      "grad_norm": 96.62998962402344,
      "learning_rate": 1.1011138846811448e-06,
      "loss": 0.7572,
      "step": 39
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 538.8275756835938,
      "learning_rate": 1.0992186800603509e-06,
      "loss": 30.8125,
      "step": 40
    },
    {
      "epoch": 0.33064516129032256,
      "grad_norm": 285.717041015625,
      "learning_rate": 1.0973234754395574e-06,
      "loss": 1.837,
      "step": 41
    },
    {
      "epoch": 0.3387096774193548,
      "grad_norm": 215.8197021484375,
      "learning_rate": 1.0954282708187636e-06,
      "loss": 18.2501,
      "step": 42
    },
    {
      "epoch": 0.3467741935483871,
      "grad_norm": 163.01434326171875,
      "learning_rate": 1.0935330661979699e-06,
      "loss": 27.9919,
      "step": 43
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 24.41200828552246,
      "learning_rate": 1.0916378615771762e-06,
      "loss": 0.3323,
      "step": 44
    },
    {
      "epoch": 0.3629032258064516,
      "grad_norm": 39.837669372558594,
      "learning_rate": 1.0897426569563824e-06,
      "loss": 1.1347,
      "step": 45
    },
    {
      "epoch": 0.3709677419354839,
      "grad_norm": 47.366092681884766,
      "learning_rate": 1.0878474523355887e-06,
      "loss": 2.3667,
      "step": 46
    },
    {
      "epoch": 0.3790322580645161,
      "grad_norm": 85.20360565185547,
      "learning_rate": 1.085952247714795e-06,
      "loss": 2.8234,
      "step": 47
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 209.84487915039062,
      "learning_rate": 1.0840570430940014e-06,
      "loss": 2.7326,
      "step": 48
    },
    {
      "epoch": 0.3951612903225806,
      "grad_norm": 61.69172286987305,
      "learning_rate": 1.0821618384732077e-06,
      "loss": 0.9325,
      "step": 49
    },
    {
      "epoch": 0.4032258064516129,
      "grad_norm": 149.46607971191406,
      "learning_rate": 1.080266633852414e-06,
      "loss": 7.6034,
      "step": 50
    },
    {
      "epoch": 0.4112903225806452,
      "grad_norm": 163.2115478515625,
      "learning_rate": 1.0783714292316202e-06,
      "loss": 1.8005,
      "step": 51
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 42.81549072265625,
      "learning_rate": 1.0764762246108265e-06,
      "loss": 0.3282,
      "step": 52
    },
    {
      "epoch": 0.4274193548387097,
      "grad_norm": 516.490966796875,
      "learning_rate": 1.0745810199900328e-06,
      "loss": 13.6021,
      "step": 53
    },
    {
      "epoch": 0.43548387096774194,
      "grad_norm": 83.2822265625,
      "learning_rate": 1.072685815369239e-06,
      "loss": 1.2216,
      "step": 54
    },
    {
      "epoch": 0.4435483870967742,
      "grad_norm": 86.0993881225586,
      "learning_rate": 1.0707906107484453e-06,
      "loss": 4.7606,
      "step": 55
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 44.0859375,
      "learning_rate": 1.0688954061276516e-06,
      "loss": 0.8722,
      "step": 56
    },
    {
      "epoch": 0.4596774193548387,
      "grad_norm": 124.6083984375,
      "learning_rate": 1.067000201506858e-06,
      "loss": 1.3964,
      "step": 57
    },
    {
      "epoch": 0.46774193548387094,
      "grad_norm": 206.24273681640625,
      "learning_rate": 1.0651049968860643e-06,
      "loss": 7.95,
      "step": 58
    },
    {
      "epoch": 0.47580645161290325,
      "grad_norm": 75.12397766113281,
      "learning_rate": 1.0632097922652704e-06,
      "loss": 0.7911,
      "step": 59
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 234.4656524658203,
      "learning_rate": 1.0613145876444769e-06,
      "loss": 14.9343,
      "step": 60
    },
    {
      "epoch": 0.49193548387096775,
      "grad_norm": 244.2675323486328,
      "learning_rate": 1.0594193830236831e-06,
      "loss": 0.5364,
      "step": 61
    },
    {
      "epoch": 0.5,
      "grad_norm": 134.83209228515625,
      "learning_rate": 1.0575241784028894e-06,
      "loss": 23.3423,
      "step": 62
    },
    {
      "epoch": 0.5080645161290323,
      "grad_norm": 117.73362731933594,
      "learning_rate": 1.0556289737820957e-06,
      "loss": 1.3623,
      "step": 63
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 67.53313446044922,
      "learning_rate": 1.053733769161302e-06,
      "loss": 2.8317,
      "step": 64
    },
    {
      "epoch": 0.5241935483870968,
      "grad_norm": 169.532470703125,
      "learning_rate": 1.0518385645405082e-06,
      "loss": 14.4335,
      "step": 65
    },
    {
      "epoch": 0.532258064516129,
      "grad_norm": 62.2873649597168,
      "learning_rate": 1.0499433599197147e-06,
      "loss": 1.3316,
      "step": 66
    },
    {
      "epoch": 0.5403225806451613,
      "grad_norm": 70.10298156738281,
      "learning_rate": 1.048048155298921e-06,
      "loss": 1.325,
      "step": 67
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 60.70389175415039,
      "learning_rate": 1.0461529506781272e-06,
      "loss": 0.8161,
      "step": 68
    },
    {
      "epoch": 0.5564516129032258,
      "grad_norm": 104.86827087402344,
      "learning_rate": 1.0442577460573335e-06,
      "loss": 1.3411,
      "step": 69
    },
    {
      "epoch": 0.5645161290322581,
      "grad_norm": 78.47552490234375,
      "learning_rate": 1.0423625414365397e-06,
      "loss": 0.7227,
      "step": 70
    },
    {
      "epoch": 0.5725806451612904,
      "grad_norm": 74.83175659179688,
      "learning_rate": 1.040467336815746e-06,
      "loss": 2.0944,
      "step": 71
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 261.87298583984375,
      "learning_rate": 1.0385721321949523e-06,
      "loss": 17.7856,
      "step": 72
    },
    {
      "epoch": 0.5887096774193549,
      "grad_norm": 55.27876663208008,
      "learning_rate": 1.0366769275741585e-06,
      "loss": 0.98,
      "step": 73
    },
    {
      "epoch": 0.5967741935483871,
      "grad_norm": 145.73300170898438,
      "learning_rate": 1.0347817229533648e-06,
      "loss": 1.7042,
      "step": 74
    },
    {
      "epoch": 0.6048387096774194,
      "grad_norm": 125.30365753173828,
      "learning_rate": 1.0328865183325713e-06,
      "loss": 16.882,
      "step": 75
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 26.542724609375,
      "learning_rate": 1.0309913137117776e-06,
      "loss": 0.1877,
      "step": 76
    },
    {
      "epoch": 0.6209677419354839,
      "grad_norm": 37.75946044921875,
      "learning_rate": 1.0290961090909838e-06,
      "loss": 0.2059,
      "step": 77
    },
    {
      "epoch": 0.6290322580645161,
      "grad_norm": 128.78330993652344,
      "learning_rate": 1.02720090447019e-06,
      "loss": 13.4677,
      "step": 78
    },
    {
      "epoch": 0.6370967741935484,
      "grad_norm": 210.26707458496094,
      "learning_rate": 1.0253056998493964e-06,
      "loss": 22.7431,
      "step": 79
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 35.724334716796875,
      "learning_rate": 1.0234104952286026e-06,
      "loss": 1.0135,
      "step": 80
    },
    {
      "epoch": 0.6532258064516129,
      "grad_norm": 66.4817886352539,
      "learning_rate": 1.0215152906078089e-06,
      "loss": 0.4585,
      "step": 81
    },
    {
      "epoch": 0.6612903225806451,
      "grad_norm": 53.807220458984375,
      "learning_rate": 1.0196200859870154e-06,
      "loss": 2.1388,
      "step": 82
    },
    {
      "epoch": 0.6693548387096774,
      "grad_norm": 33.74148178100586,
      "learning_rate": 1.0177248813662214e-06,
      "loss": 0.2365,
      "step": 83
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 47.4416618347168,
      "learning_rate": 1.015829676745428e-06,
      "loss": 1.1393,
      "step": 84
    },
    {
      "epoch": 0.6854838709677419,
      "grad_norm": 280.392578125,
      "learning_rate": 1.0139344721246342e-06,
      "loss": 18.5951,
      "step": 85
    },
    {
      "epoch": 0.6935483870967742,
      "grad_norm": 91.39240264892578,
      "learning_rate": 1.0120392675038404e-06,
      "loss": 14.467,
      "step": 86
    },
    {
      "epoch": 0.7016129032258065,
      "grad_norm": 97.39054107666016,
      "learning_rate": 1.0101440628830467e-06,
      "loss": 2.257,
      "step": 87
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 95.57569885253906,
      "learning_rate": 1.008248858262253e-06,
      "loss": 0.9064,
      "step": 88
    },
    {
      "epoch": 0.717741935483871,
      "grad_norm": 230.4451904296875,
      "learning_rate": 1.0063536536414592e-06,
      "loss": 1.5809,
      "step": 89
    },
    {
      "epoch": 0.7258064516129032,
      "grad_norm": 149.36996459960938,
      "learning_rate": 1.0044584490206655e-06,
      "loss": 1.7855,
      "step": 90
    },
    {
      "epoch": 0.7338709677419355,
      "grad_norm": 319.1463317871094,
      "learning_rate": 1.002563244399872e-06,
      "loss": 19.9728,
      "step": 91
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 132.4302215576172,
      "learning_rate": 1.000668039779078e-06,
      "loss": 7.4546,
      "step": 92
    },
    {
      "epoch": 0.75,
      "grad_norm": 24.757253646850586,
      "learning_rate": 9.987728351582845e-07,
      "loss": 0.1834,
      "step": 93
    },
    {
      "epoch": 0.7580645161290323,
      "grad_norm": 117.05673217773438,
      "learning_rate": 9.968776305374908e-07,
      "loss": 1.7553,
      "step": 94
    },
    {
      "epoch": 0.7661290322580645,
      "grad_norm": 178.41818237304688,
      "learning_rate": 9.94982425916697e-07,
      "loss": 9.5775,
      "step": 95
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 129.0936737060547,
      "learning_rate": 9.930872212959033e-07,
      "loss": 2.2522,
      "step": 96
    },
    {
      "epoch": 0.782258064516129,
      "grad_norm": 120.8997802734375,
      "learning_rate": 9.911920166751096e-07,
      "loss": 1.365,
      "step": 97
    },
    {
      "epoch": 0.7903225806451613,
      "grad_norm": 124.6778335571289,
      "learning_rate": 9.892968120543159e-07,
      "loss": 1.6905,
      "step": 98
    },
    {
      "epoch": 0.7983870967741935,
      "grad_norm": 805.3359375,
      "learning_rate": 9.874016074335221e-07,
      "loss": 2.0304,
      "step": 99
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 258.4070739746094,
      "learning_rate": 9.855064028127286e-07,
      "loss": 18.573,
      "step": 100
    },
    {
      "epoch": 0.8145161290322581,
      "grad_norm": 25.10341453552246,
      "learning_rate": 9.836111981919347e-07,
      "loss": 0.8904,
      "step": 101
    },
    {
      "epoch": 0.8225806451612904,
      "grad_norm": 137.26222229003906,
      "learning_rate": 9.817159935711411e-07,
      "loss": 1.3759,
      "step": 102
    },
    {
      "epoch": 0.8306451612903226,
      "grad_norm": 100.84027862548828,
      "learning_rate": 9.798207889503474e-07,
      "loss": 11.2571,
      "step": 103
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 99.84913635253906,
      "learning_rate": 9.779255843295537e-07,
      "loss": 0.7655,
      "step": 104
    },
    {
      "epoch": 0.8467741935483871,
      "grad_norm": 58.0277099609375,
      "learning_rate": 9.7603037970876e-07,
      "loss": 0.685,
      "step": 105
    },
    {
      "epoch": 0.8548387096774194,
      "grad_norm": 118.5246810913086,
      "learning_rate": 9.741351750879662e-07,
      "loss": 3.1796,
      "step": 106
    },
    {
      "epoch": 0.8629032258064516,
      "grad_norm": 93.28719329833984,
      "learning_rate": 9.722399704671725e-07,
      "loss": 1.5221,
      "step": 107
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 109.0409927368164,
      "learning_rate": 9.703447658463787e-07,
      "loss": 1.2383,
      "step": 108
    },
    {
      "epoch": 0.8790322580645161,
      "grad_norm": 167.7772216796875,
      "learning_rate": 9.684495612255852e-07,
      "loss": 2.5045,
      "step": 109
    },
    {
      "epoch": 0.8870967741935484,
      "grad_norm": 51.926692962646484,
      "learning_rate": 9.665543566047915e-07,
      "loss": 1.0717,
      "step": 110
    },
    {
      "epoch": 0.8951612903225806,
      "grad_norm": 222.6824951171875,
      "learning_rate": 9.646591519839975e-07,
      "loss": 4.1007,
      "step": 111
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 53.584320068359375,
      "learning_rate": 9.62763947363204e-07,
      "loss": 1.257,
      "step": 112
    },
    {
      "epoch": 0.9112903225806451,
      "grad_norm": 92.6733627319336,
      "learning_rate": 9.608687427424103e-07,
      "loss": 0.7563,
      "step": 113
    },
    {
      "epoch": 0.9193548387096774,
      "grad_norm": 23.77389144897461,
      "learning_rate": 9.589735381216166e-07,
      "loss": 0.7985,
      "step": 114
    },
    {
      "epoch": 0.9274193548387096,
      "grad_norm": 374.8204650878906,
      "learning_rate": 9.570783335008228e-07,
      "loss": 5.5011,
      "step": 115
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 83.84817504882812,
      "learning_rate": 9.55183128880029e-07,
      "loss": 2.123,
      "step": 116
    },
    {
      "epoch": 0.9435483870967742,
      "grad_norm": 160.6682891845703,
      "learning_rate": 9.532879242592355e-07,
      "loss": 1.6017,
      "step": 117
    },
    {
      "epoch": 0.9516129032258065,
      "grad_norm": 150.16294860839844,
      "learning_rate": 9.513927196384417e-07,
      "loss": 14.0088,
      "step": 118
    },
    {
      "epoch": 0.9596774193548387,
      "grad_norm": 94.02184295654297,
      "learning_rate": 9.494975150176481e-07,
      "loss": 1.6218,
      "step": 119
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 40.898277282714844,
      "learning_rate": 9.476023103968543e-07,
      "loss": 1.4986,
      "step": 120
    },
    {
      "epoch": 0.9758064516129032,
      "grad_norm": 167.00746154785156,
      "learning_rate": 9.457071057760605e-07,
      "loss": 1.4057,
      "step": 121
    },
    {
      "epoch": 0.9838709677419355,
      "grad_norm": 35.20161819458008,
      "learning_rate": 9.438119011552669e-07,
      "loss": 0.5185,
      "step": 122
    },
    {
      "epoch": 0.9919354838709677,
      "grad_norm": 61.312583923339844,
      "learning_rate": 9.419166965344732e-07,
      "loss": 0.6058,
      "step": 123
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.15098762512207,
      "learning_rate": 9.400214919136795e-07,
      "loss": 0.008,
      "step": 124
    },
    {
      "epoch": 1.0,
      "eval_loss": 3.5903782844543457,
      "eval_mae": 1.1498976945877075,
      "eval_mse": 3.590378522872925,
      "eval_r2": 0.01701676845550537,
      "eval_rmse": 1.8948294178824976,
      "eval_runtime": 1.3414,
      "eval_samples_per_second": 41.002,
      "eval_smape": 46.445611119270325,
      "eval_steps_per_second": 10.437,
      "step": 124
    },
    {
      "epoch": 1.0080645161290323,
      "grad_norm": 93.99532318115234,
      "learning_rate": 9.381262872928857e-07,
      "loss": 1.3843,
      "step": 125
    },
    {
      "epoch": 1.0161290322580645,
      "grad_norm": 131.81895446777344,
      "learning_rate": 9.362310826720921e-07,
      "loss": 0.7999,
      "step": 126
    },
    {
      "epoch": 1.0241935483870968,
      "grad_norm": 137.10763549804688,
      "learning_rate": 9.343358780512983e-07,
      "loss": 6.5196,
      "step": 127
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 48.18350601196289,
      "learning_rate": 9.324406734305047e-07,
      "loss": 0.6827,
      "step": 128
    },
    {
      "epoch": 1.0403225806451613,
      "grad_norm": 177.0278778076172,
      "learning_rate": 9.30545468809711e-07,
      "loss": 0.8596,
      "step": 129
    },
    {
      "epoch": 1.0483870967741935,
      "grad_norm": 47.4871711730957,
      "learning_rate": 9.286502641889171e-07,
      "loss": 2.1029,
      "step": 130
    },
    {
      "epoch": 1.0564516129032258,
      "grad_norm": 47.476219177246094,
      "learning_rate": 9.267550595681235e-07,
      "loss": 0.2775,
      "step": 131
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 171.87510681152344,
      "learning_rate": 9.248598549473298e-07,
      "loss": 5.2664,
      "step": 132
    },
    {
      "epoch": 1.0725806451612903,
      "grad_norm": 76.38048553466797,
      "learning_rate": 9.229646503265362e-07,
      "loss": 0.9048,
      "step": 133
    },
    {
      "epoch": 1.0806451612903225,
      "grad_norm": 65.83001708984375,
      "learning_rate": 9.210694457057423e-07,
      "loss": 0.6209,
      "step": 134
    },
    {
      "epoch": 1.0887096774193548,
      "grad_norm": 155.04318237304688,
      "learning_rate": 9.191742410849487e-07,
      "loss": 7.5308,
      "step": 135
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 180.7427520751953,
      "learning_rate": 9.17279036464155e-07,
      "loss": 16.1446,
      "step": 136
    },
    {
      "epoch": 1.1048387096774193,
      "grad_norm": 209.10556030273438,
      "learning_rate": 9.153838318433613e-07,
      "loss": 10.6327,
      "step": 137
    },
    {
      "epoch": 1.1129032258064515,
      "grad_norm": 31.048919677734375,
      "learning_rate": 9.134886272225676e-07,
      "loss": 0.4128,
      "step": 138
    },
    {
      "epoch": 1.120967741935484,
      "grad_norm": 419.14520263671875,
      "learning_rate": 9.115934226017738e-07,
      "loss": 45.5376,
      "step": 139
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 18.564205169677734,
      "learning_rate": 9.096982179809801e-07,
      "loss": 0.1034,
      "step": 140
    },
    {
      "epoch": 1.1370967741935485,
      "grad_norm": 38.12897872924805,
      "learning_rate": 9.078030133601864e-07,
      "loss": 1.2526,
      "step": 141
    },
    {
      "epoch": 1.1451612903225807,
      "grad_norm": 56.46409606933594,
      "learning_rate": 9.059078087393928e-07,
      "loss": 1.3236,
      "step": 142
    },
    {
      "epoch": 1.153225806451613,
      "grad_norm": 57.76015090942383,
      "learning_rate": 9.04012604118599e-07,
      "loss": 0.3464,
      "step": 143
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 32.81700134277344,
      "learning_rate": 9.021173994978053e-07,
      "loss": 0.2656,
      "step": 144
    },
    {
      "epoch": 1.1693548387096775,
      "grad_norm": 40.10895919799805,
      "learning_rate": 9.002221948770116e-07,
      "loss": 0.8185,
      "step": 145
    },
    {
      "epoch": 1.1774193548387097,
      "grad_norm": 46.941200256347656,
      "learning_rate": 8.98326990256218e-07,
      "loss": 2.0041,
      "step": 146
    },
    {
      "epoch": 1.185483870967742,
      "grad_norm": 47.112571716308594,
      "learning_rate": 8.964317856354242e-07,
      "loss": 0.4361,
      "step": 147
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 91.99717712402344,
      "learning_rate": 8.945365810146304e-07,
      "loss": 1.2795,
      "step": 148
    },
    {
      "epoch": 1.2016129032258065,
      "grad_norm": 321.48345947265625,
      "learning_rate": 8.926413763938367e-07,
      "loss": 21.7397,
      "step": 149
    },
    {
      "epoch": 1.2096774193548387,
      "grad_norm": 420.6986083984375,
      "learning_rate": 8.90746171773043e-07,
      "loss": 37.258,
      "step": 150
    },
    {
      "epoch": 1.217741935483871,
      "grad_norm": 195.8186492919922,
      "learning_rate": 8.888509671522494e-07,
      "loss": 17.6255,
      "step": 151
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 131.71688842773438,
      "learning_rate": 8.869557625314557e-07,
      "loss": 1.9709,
      "step": 152
    },
    {
      "epoch": 1.2338709677419355,
      "grad_norm": 275.066162109375,
      "learning_rate": 8.850605579106619e-07,
      "loss": 10.5323,
      "step": 153
    },
    {
      "epoch": 1.2419354838709677,
      "grad_norm": 121.62479400634766,
      "learning_rate": 8.831653532898682e-07,
      "loss": 13.0309,
      "step": 154
    },
    {
      "epoch": 1.25,
      "grad_norm": 44.131412506103516,
      "learning_rate": 8.812701486690746e-07,
      "loss": 0.4236,
      "step": 155
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 48.05416488647461,
      "learning_rate": 8.793749440482808e-07,
      "loss": 0.787,
      "step": 156
    },
    {
      "epoch": 1.2661290322580645,
      "grad_norm": 136.1278839111328,
      "learning_rate": 8.774797394274871e-07,
      "loss": 4.653,
      "step": 157
    },
    {
      "epoch": 1.2741935483870968,
      "grad_norm": 156.81910705566406,
      "learning_rate": 8.755845348066934e-07,
      "loss": 14.6158,
      "step": 158
    },
    {
      "epoch": 1.282258064516129,
      "grad_norm": 311.3755798339844,
      "learning_rate": 8.736893301858996e-07,
      "loss": 25.6107,
      "step": 159
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 147.80181884765625,
      "learning_rate": 8.71794125565106e-07,
      "loss": 10.7386,
      "step": 160
    },
    {
      "epoch": 1.2983870967741935,
      "grad_norm": 326.3906555175781,
      "learning_rate": 8.698989209443123e-07,
      "loss": 41.9267,
      "step": 161
    },
    {
      "epoch": 1.3064516129032258,
      "grad_norm": 35.69755172729492,
      "learning_rate": 8.680037163235186e-07,
      "loss": 0.9466,
      "step": 162
    },
    {
      "epoch": 1.314516129032258,
      "grad_norm": 82.98596954345703,
      "learning_rate": 8.661085117027248e-07,
      "loss": 0.6258,
      "step": 163
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 105.8860855102539,
      "learning_rate": 8.642133070819311e-07,
      "loss": 2.386,
      "step": 164
    },
    {
      "epoch": 1.3306451612903225,
      "grad_norm": 196.2127685546875,
      "learning_rate": 8.623181024611374e-07,
      "loss": 2.511,
      "step": 165
    },
    {
      "epoch": 1.3387096774193548,
      "grad_norm": 145.88681030273438,
      "learning_rate": 8.604228978403437e-07,
      "loss": 1.8581,
      "step": 166
    },
    {
      "epoch": 1.346774193548387,
      "grad_norm": 151.8137664794922,
      "learning_rate": 8.5852769321955e-07,
      "loss": 3.5077,
      "step": 167
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 83.18711853027344,
      "learning_rate": 8.566324885987562e-07,
      "loss": 0.5735,
      "step": 168
    },
    {
      "epoch": 1.3629032258064515,
      "grad_norm": 137.17172241210938,
      "learning_rate": 8.547372839779626e-07,
      "loss": 2.0364,
      "step": 169
    },
    {
      "epoch": 1.370967741935484,
      "grad_norm": 486.7686767578125,
      "learning_rate": 8.528420793571689e-07,
      "loss": 130.1139,
      "step": 170
    },
    {
      "epoch": 1.379032258064516,
      "grad_norm": 53.59453582763672,
      "learning_rate": 8.509468747363753e-07,
      "loss": 1.0157,
      "step": 171
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 361.83270263671875,
      "learning_rate": 8.490516701155814e-07,
      "loss": 9.2801,
      "step": 172
    },
    {
      "epoch": 1.3951612903225805,
      "grad_norm": 122.44477081298828,
      "learning_rate": 8.471564654947877e-07,
      "loss": 1.7689,
      "step": 173
    },
    {
      "epoch": 1.403225806451613,
      "grad_norm": 163.0518035888672,
      "learning_rate": 8.452612608739941e-07,
      "loss": 8.8246,
      "step": 174
    },
    {
      "epoch": 1.4112903225806452,
      "grad_norm": 23.73016357421875,
      "learning_rate": 8.433660562532003e-07,
      "loss": 0.3707,
      "step": 175
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 392.39739990234375,
      "learning_rate": 8.414708516324067e-07,
      "loss": 19.7684,
      "step": 176
    },
    {
      "epoch": 1.4274193548387097,
      "grad_norm": 101.17041015625,
      "learning_rate": 8.395756470116129e-07,
      "loss": 1.2123,
      "step": 177
    },
    {
      "epoch": 1.435483870967742,
      "grad_norm": 129.82452392578125,
      "learning_rate": 8.376804423908192e-07,
      "loss": 1.3362,
      "step": 178
    },
    {
      "epoch": 1.4435483870967742,
      "grad_norm": 93.25836944580078,
      "learning_rate": 8.357852377700255e-07,
      "loss": 2.67,
      "step": 179
    },
    {
      "epoch": 1.4516129032258065,
      "grad_norm": 144.9728240966797,
      "learning_rate": 8.338900331492319e-07,
      "loss": 1.4356,
      "step": 180
    },
    {
      "epoch": 1.4596774193548387,
      "grad_norm": 106.55712127685547,
      "learning_rate": 8.31994828528438e-07,
      "loss": 1.0411,
      "step": 181
    },
    {
      "epoch": 1.467741935483871,
      "grad_norm": 50.07807922363281,
      "learning_rate": 8.300996239076443e-07,
      "loss": 1.0384,
      "step": 182
    },
    {
      "epoch": 1.4758064516129032,
      "grad_norm": 47.749080657958984,
      "learning_rate": 8.282044192868507e-07,
      "loss": 0.8744,
      "step": 183
    },
    {
      "epoch": 1.4838709677419355,
      "grad_norm": 505.6190185546875,
      "learning_rate": 8.263092146660569e-07,
      "loss": 3.8061,
      "step": 184
    },
    {
      "epoch": 1.4919354838709677,
      "grad_norm": 126.373779296875,
      "learning_rate": 8.244140100452633e-07,
      "loss": 1.3906,
      "step": 185
    },
    {
      "epoch": 1.5,
      "grad_norm": 95.14514923095703,
      "learning_rate": 8.225188054244695e-07,
      "loss": 0.7666,
      "step": 186
    },
    {
      "epoch": 1.5080645161290323,
      "grad_norm": 171.452392578125,
      "learning_rate": 8.206236008036759e-07,
      "loss": 4.4224,
      "step": 187
    },
    {
      "epoch": 1.5161290322580645,
      "grad_norm": 58.15391159057617,
      "learning_rate": 8.187283961828821e-07,
      "loss": 2.4317,
      "step": 188
    },
    {
      "epoch": 1.5241935483870968,
      "grad_norm": 73.09158325195312,
      "learning_rate": 8.168331915620885e-07,
      "loss": 1.3731,
      "step": 189
    },
    {
      "epoch": 1.532258064516129,
      "grad_norm": 92.6122817993164,
      "learning_rate": 8.149379869412948e-07,
      "loss": 1.0538,
      "step": 190
    },
    {
      "epoch": 1.5403225806451613,
      "grad_norm": 61.34569549560547,
      "learning_rate": 8.130427823205009e-07,
      "loss": 0.8926,
      "step": 191
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 46.87858200073242,
      "learning_rate": 8.111475776997073e-07,
      "loss": 0.774,
      "step": 192
    },
    {
      "epoch": 1.5564516129032258,
      "grad_norm": 32.91598129272461,
      "learning_rate": 8.092523730789136e-07,
      "loss": 0.6749,
      "step": 193
    },
    {
      "epoch": 1.564516129032258,
      "grad_norm": 69.08930206298828,
      "learning_rate": 8.073571684581199e-07,
      "loss": 0.7667,
      "step": 194
    },
    {
      "epoch": 1.5725806451612905,
      "grad_norm": 61.4034538269043,
      "learning_rate": 8.054619638373261e-07,
      "loss": 0.6325,
      "step": 195
    },
    {
      "epoch": 1.5806451612903225,
      "grad_norm": 252.8091583251953,
      "learning_rate": 8.035667592165325e-07,
      "loss": 9.2466,
      "step": 196
    },
    {
      "epoch": 1.588709677419355,
      "grad_norm": 60.166046142578125,
      "learning_rate": 8.016715545957387e-07,
      "loss": 1.7571,
      "step": 197
    },
    {
      "epoch": 1.596774193548387,
      "grad_norm": 42.4896125793457,
      "learning_rate": 7.997763499749451e-07,
      "loss": 0.5738,
      "step": 198
    },
    {
      "epoch": 1.6048387096774195,
      "grad_norm": 62.25552749633789,
      "learning_rate": 7.978811453541514e-07,
      "loss": 0.356,
      "step": 199
    },
    {
      "epoch": 1.6129032258064515,
      "grad_norm": 131.12069702148438,
      "learning_rate": 7.959859407333575e-07,
      "loss": 1.0522,
      "step": 200
    },
    {
      "epoch": 1.620967741935484,
      "grad_norm": 13.091470718383789,
      "learning_rate": 7.940907361125639e-07,
      "loss": 0.2088,
      "step": 201
    },
    {
      "epoch": 1.629032258064516,
      "grad_norm": 61.03318786621094,
      "learning_rate": 7.921955314917702e-07,
      "loss": 1.6789,
      "step": 202
    },
    {
      "epoch": 1.6370967741935485,
      "grad_norm": 14.173700332641602,
      "learning_rate": 7.903003268709766e-07,
      "loss": 0.1112,
      "step": 203
    },
    {
      "epoch": 1.6451612903225805,
      "grad_norm": 70.016845703125,
      "learning_rate": 7.884051222501828e-07,
      "loss": 0.8495,
      "step": 204
    },
    {
      "epoch": 1.653225806451613,
      "grad_norm": 217.45611572265625,
      "learning_rate": 7.865099176293891e-07,
      "loss": 15.7051,
      "step": 205
    },
    {
      "epoch": 1.661290322580645,
      "grad_norm": 99.29098510742188,
      "learning_rate": 7.846147130085954e-07,
      "loss": 1.0348,
      "step": 206
    },
    {
      "epoch": 1.6693548387096775,
      "grad_norm": 33.91508102416992,
      "learning_rate": 7.827195083878016e-07,
      "loss": 0.7734,
      "step": 207
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 93.15892028808594,
      "learning_rate": 7.80824303767008e-07,
      "loss": 0.3326,
      "step": 208
    },
    {
      "epoch": 1.685483870967742,
      "grad_norm": 14.714130401611328,
      "learning_rate": 7.789290991462143e-07,
      "loss": 0.3946,
      "step": 209
    },
    {
      "epoch": 1.6935483870967742,
      "grad_norm": 22.98834800720215,
      "learning_rate": 7.770338945254205e-07,
      "loss": 0.2196,
      "step": 210
    },
    {
      "epoch": 1.7016129032258065,
      "grad_norm": 165.18626403808594,
      "learning_rate": 7.751386899046268e-07,
      "loss": 3.0653,
      "step": 211
    },
    {
      "epoch": 1.7096774193548387,
      "grad_norm": 55.49916458129883,
      "learning_rate": 7.732434852838332e-07,
      "loss": 0.8254,
      "step": 212
    },
    {
      "epoch": 1.717741935483871,
      "grad_norm": 25.563867568969727,
      "learning_rate": 7.713482806630394e-07,
      "loss": 0.1924,
      "step": 213
    },
    {
      "epoch": 1.7258064516129032,
      "grad_norm": 277.5033874511719,
      "learning_rate": 7.694530760422456e-07,
      "loss": 26.1251,
      "step": 214
    },
    {
      "epoch": 1.7338709677419355,
      "grad_norm": 218.53463745117188,
      "learning_rate": 7.67557871421452e-07,
      "loss": 16.69,
      "step": 215
    },
    {
      "epoch": 1.7419354838709677,
      "grad_norm": 48.32767868041992,
      "learning_rate": 7.656626668006582e-07,
      "loss": 1.267,
      "step": 216
    },
    {
      "epoch": 1.75,
      "grad_norm": 297.5074462890625,
      "learning_rate": 7.637674621798646e-07,
      "loss": 8.8952,
      "step": 217
    },
    {
      "epoch": 1.7580645161290323,
      "grad_norm": 19.5325927734375,
      "learning_rate": 7.618722575590709e-07,
      "loss": 0.1047,
      "step": 218
    },
    {
      "epoch": 1.7661290322580645,
      "grad_norm": 382.64434814453125,
      "learning_rate": 7.599770529382771e-07,
      "loss": 18.6839,
      "step": 219
    },
    {
      "epoch": 1.7741935483870968,
      "grad_norm": 18.517972946166992,
      "learning_rate": 7.580818483174834e-07,
      "loss": 0.1679,
      "step": 220
    },
    {
      "epoch": 1.782258064516129,
      "grad_norm": 36.31782913208008,
      "learning_rate": 7.561866436966898e-07,
      "loss": 0.1592,
      "step": 221
    },
    {
      "epoch": 1.7903225806451613,
      "grad_norm": 44.47116470336914,
      "learning_rate": 7.54291439075896e-07,
      "loss": 0.6091,
      "step": 222
    },
    {
      "epoch": 1.7983870967741935,
      "grad_norm": 16.11575698852539,
      "learning_rate": 7.523962344551024e-07,
      "loss": 0.1174,
      "step": 223
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 288.2611999511719,
      "learning_rate": 7.505010298343086e-07,
      "loss": 18.947,
      "step": 224
    },
    {
      "epoch": 1.814516129032258,
      "grad_norm": 267.47540283203125,
      "learning_rate": 7.486058252135149e-07,
      "loss": 10.6328,
      "step": 225
    },
    {
      "epoch": 1.8225806451612905,
      "grad_norm": 33.86650848388672,
      "learning_rate": 7.467106205927212e-07,
      "loss": 2.1226,
      "step": 226
    },
    {
      "epoch": 1.8306451612903225,
      "grad_norm": 121.6884765625,
      "learning_rate": 7.448154159719275e-07,
      "loss": 0.371,
      "step": 227
    },
    {
      "epoch": 1.838709677419355,
      "grad_norm": 42.18556594848633,
      "learning_rate": 7.429202113511338e-07,
      "loss": 0.7408,
      "step": 228
    },
    {
      "epoch": 1.846774193548387,
      "grad_norm": 64.0304946899414,
      "learning_rate": 7.4102500673034e-07,
      "loss": 0.7927,
      "step": 229
    },
    {
      "epoch": 1.8548387096774195,
      "grad_norm": 211.26959228515625,
      "learning_rate": 7.391298021095464e-07,
      "loss": 8.6268,
      "step": 230
    },
    {
      "epoch": 1.8629032258064515,
      "grad_norm": 97.74939727783203,
      "learning_rate": 7.372345974887527e-07,
      "loss": 1.1381,
      "step": 231
    },
    {
      "epoch": 1.870967741935484,
      "grad_norm": 325.8834228515625,
      "learning_rate": 7.35339392867959e-07,
      "loss": 1.0786,
      "step": 232
    },
    {
      "epoch": 1.879032258064516,
      "grad_norm": 49.10792541503906,
      "learning_rate": 7.334441882471652e-07,
      "loss": 0.5383,
      "step": 233
    },
    {
      "epoch": 1.8870967741935485,
      "grad_norm": 27.410547256469727,
      "learning_rate": 7.315489836263715e-07,
      "loss": 0.4044,
      "step": 234
    },
    {
      "epoch": 1.8951612903225805,
      "grad_norm": 31.08856773376465,
      "learning_rate": 7.296537790055778e-07,
      "loss": 0.2857,
      "step": 235
    },
    {
      "epoch": 1.903225806451613,
      "grad_norm": 236.7850341796875,
      "learning_rate": 7.277585743847841e-07,
      "loss": 10.882,
      "step": 236
    },
    {
      "epoch": 1.911290322580645,
      "grad_norm": 85.23385620117188,
      "learning_rate": 7.258633697639905e-07,
      "loss": 6.385,
      "step": 237
    },
    {
      "epoch": 1.9193548387096775,
      "grad_norm": 42.652278900146484,
      "learning_rate": 7.239681651431966e-07,
      "loss": 0.7745,
      "step": 238
    },
    {
      "epoch": 1.9274193548387095,
      "grad_norm": 37.44199752807617,
      "learning_rate": 7.22072960522403e-07,
      "loss": 0.6186,
      "step": 239
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 122.3661117553711,
      "learning_rate": 7.201777559016093e-07,
      "loss": 3.3579,
      "step": 240
    },
    {
      "epoch": 1.9435483870967742,
      "grad_norm": 29.615245819091797,
      "learning_rate": 7.182825512808157e-07,
      "loss": 1.0682,
      "step": 241
    },
    {
      "epoch": 1.9516129032258065,
      "grad_norm": 38.644981384277344,
      "learning_rate": 7.163873466600218e-07,
      "loss": 1.6679,
      "step": 242
    },
    {
      "epoch": 1.9596774193548387,
      "grad_norm": 151.14231872558594,
      "learning_rate": 7.144921420392281e-07,
      "loss": 3.0028,
      "step": 243
    },
    {
      "epoch": 1.967741935483871,
      "grad_norm": 90.12117767333984,
      "learning_rate": 7.125969374184345e-07,
      "loss": 3.1295,
      "step": 244
    },
    {
      "epoch": 1.9758064516129032,
      "grad_norm": 65.80992889404297,
      "learning_rate": 7.107017327976407e-07,
      "loss": 0.8262,
      "step": 245
    },
    {
      "epoch": 1.9838709677419355,
      "grad_norm": 224.11376953125,
      "learning_rate": 7.088065281768471e-07,
      "loss": 2.9559,
      "step": 246
    },
    {
      "epoch": 1.9919354838709677,
      "grad_norm": 165.00535583496094,
      "learning_rate": 7.069113235560533e-07,
      "loss": 1.3005,
      "step": 247
    },
    {
      "epoch": 2.0,
      "grad_norm": 178.8065185546875,
      "learning_rate": 7.050161189352596e-07,
      "loss": 2.704,
      "step": 248
    },
    {
      "epoch": 2.0,
      "eval_loss": 3.476463556289673,
      "eval_mae": 1.1249998807907104,
      "eval_mse": 3.476463794708252,
      "eval_r2": 0.04820472002029419,
      "eval_rmse": 1.8645277672129885,
      "eval_runtime": 1.4258,
      "eval_samples_per_second": 38.574,
      "eval_smape": 47.31981158256531,
      "eval_steps_per_second": 9.819,
      "step": 248
    }
  ],
  "logging_steps": 1,
  "max_steps": 620,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1343939627507712.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.1750268648920993e-06
  }
}