{
  "best_metric": 10.305556297302246,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 3.0125786163522013,
  "eval_steps": 50,
  "global_step": 119,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025157232704402517,
      "grad_norm": 0.1715589165687561,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 10.3802,
      "step": 1
    },
    {
      "epoch": 0.025157232704402517,
      "eval_loss": 10.382518768310547,
      "eval_runtime": 0.1746,
      "eval_samples_per_second": 3058.484,
      "eval_steps_per_second": 97.367,
      "step": 1
    },
    {
      "epoch": 0.050314465408805034,
      "grad_norm": 0.18782849609851837,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 10.3817,
      "step": 2
    },
    {
      "epoch": 0.07547169811320754,
      "grad_norm": 0.22927959263324738,
      "learning_rate": 3.0299999999999998e-05,
      "loss": 10.379,
      "step": 3
    },
    {
      "epoch": 0.10062893081761007,
      "grad_norm": 0.338342547416687,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 10.3812,
      "step": 4
    },
    {
      "epoch": 0.12578616352201258,
      "grad_norm": 0.34414851665496826,
      "learning_rate": 5.05e-05,
      "loss": 10.3802,
      "step": 5
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 0.46170228719711304,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 10.3843,
      "step": 6
    },
    {
      "epoch": 0.1761006289308176,
      "grad_norm": 0.1982971578836441,
      "learning_rate": 7.07e-05,
      "loss": 10.3814,
      "step": 7
    },
    {
      "epoch": 0.20125786163522014,
      "grad_norm": 0.20252886414527893,
      "learning_rate": 8.080000000000001e-05,
      "loss": 10.3824,
      "step": 8
    },
    {
      "epoch": 0.22641509433962265,
      "grad_norm": 0.23484252393245697,
      "learning_rate": 9.09e-05,
      "loss": 10.3728,
      "step": 9
    },
    {
      "epoch": 0.25157232704402516,
      "grad_norm": 0.3196582496166229,
      "learning_rate": 0.000101,
      "loss": 10.3746,
      "step": 10
    },
    {
      "epoch": 0.27672955974842767,
      "grad_norm": 0.35150107741355896,
      "learning_rate": 0.00010007339449541285,
      "loss": 10.3727,
      "step": 11
    },
    {
      "epoch": 0.3018867924528302,
      "grad_norm": 0.4719482362270355,
      "learning_rate": 9.914678899082569e-05,
      "loss": 10.3699,
      "step": 12
    },
    {
      "epoch": 0.3270440251572327,
      "grad_norm": 0.2938939034938812,
      "learning_rate": 9.822018348623854e-05,
      "loss": 10.376,
      "step": 13
    },
    {
      "epoch": 0.3522012578616352,
      "grad_norm": 0.2226066291332245,
      "learning_rate": 9.729357798165138e-05,
      "loss": 10.3751,
      "step": 14
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 0.24701771140098572,
      "learning_rate": 9.636697247706423e-05,
      "loss": 10.3662,
      "step": 15
    },
    {
      "epoch": 0.4025157232704403,
      "grad_norm": 0.3606414794921875,
      "learning_rate": 9.544036697247707e-05,
      "loss": 10.3621,
      "step": 16
    },
    {
      "epoch": 0.4276729559748428,
      "grad_norm": 0.3855360448360443,
      "learning_rate": 9.451376146788992e-05,
      "loss": 10.3593,
      "step": 17
    },
    {
      "epoch": 0.4528301886792453,
      "grad_norm": 0.4999494254589081,
      "learning_rate": 9.358715596330275e-05,
      "loss": 10.3569,
      "step": 18
    },
    {
      "epoch": 0.4779874213836478,
      "grad_norm": 0.4514361619949341,
      "learning_rate": 9.266055045871561e-05,
      "loss": 10.3608,
      "step": 19
    },
    {
      "epoch": 0.5031446540880503,
      "grad_norm": 0.26520147919654846,
      "learning_rate": 9.173394495412844e-05,
      "loss": 10.3659,
      "step": 20
    },
    {
      "epoch": 0.5283018867924528,
      "grad_norm": 0.26291126012802124,
      "learning_rate": 9.08073394495413e-05,
      "loss": 10.3592,
      "step": 21
    },
    {
      "epoch": 0.5534591194968553,
      "grad_norm": 0.36332985758781433,
      "learning_rate": 8.988073394495413e-05,
      "loss": 10.3518,
      "step": 22
    },
    {
      "epoch": 0.5786163522012578,
      "grad_norm": 0.40622854232788086,
      "learning_rate": 8.895412844036697e-05,
      "loss": 10.344,
      "step": 23
    },
    {
      "epoch": 0.6037735849056604,
      "grad_norm": 0.5071566700935364,
      "learning_rate": 8.802752293577982e-05,
      "loss": 10.3411,
      "step": 24
    },
    {
      "epoch": 0.6289308176100629,
      "grad_norm": 0.6803290843963623,
      "learning_rate": 8.710091743119266e-05,
      "loss": 10.329,
      "step": 25
    },
    {
      "epoch": 0.6540880503144654,
      "grad_norm": 0.3055112063884735,
      "learning_rate": 8.617431192660551e-05,
      "loss": 10.3552,
      "step": 26
    },
    {
      "epoch": 0.6792452830188679,
      "grad_norm": 0.2761920392513275,
      "learning_rate": 8.524770642201835e-05,
      "loss": 10.3511,
      "step": 27
    },
    {
      "epoch": 0.7044025157232704,
      "grad_norm": 0.25242385268211365,
      "learning_rate": 8.43211009174312e-05,
      "loss": 10.3422,
      "step": 28
    },
    {
      "epoch": 0.7295597484276729,
      "grad_norm": 0.35044631361961365,
      "learning_rate": 8.339449541284404e-05,
      "loss": 10.3301,
      "step": 29
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 0.4619623124599457,
      "learning_rate": 8.246788990825688e-05,
      "loss": 10.3278,
      "step": 30
    },
    {
      "epoch": 0.779874213836478,
      "grad_norm": 0.5502440333366394,
      "learning_rate": 8.154128440366973e-05,
      "loss": 10.3209,
      "step": 31
    },
    {
      "epoch": 0.8050314465408805,
      "grad_norm": 0.31955239176750183,
      "learning_rate": 8.061467889908257e-05,
      "loss": 10.3387,
      "step": 32
    },
    {
      "epoch": 0.8301886792452831,
      "grad_norm": 0.26196402311325073,
      "learning_rate": 7.968807339449542e-05,
      "loss": 10.3445,
      "step": 33
    },
    {
      "epoch": 0.8553459119496856,
      "grad_norm": 0.17065182328224182,
      "learning_rate": 7.876146788990826e-05,
      "loss": 10.337,
      "step": 34
    },
    {
      "epoch": 0.8805031446540881,
      "grad_norm": 0.28439250588417053,
      "learning_rate": 7.78348623853211e-05,
      "loss": 10.3194,
      "step": 35
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.3914329409599304,
      "learning_rate": 7.690825688073395e-05,
      "loss": 10.3172,
      "step": 36
    },
    {
      "epoch": 0.9308176100628931,
      "grad_norm": 0.4504752457141876,
      "learning_rate": 7.598165137614679e-05,
      "loss": 10.308,
      "step": 37
    },
    {
      "epoch": 0.9559748427672956,
      "grad_norm": 0.2524828612804413,
      "learning_rate": 7.505504587155964e-05,
      "loss": 10.3287,
      "step": 38
    },
    {
      "epoch": 0.9811320754716981,
      "grad_norm": 0.17636598646640778,
      "learning_rate": 7.412844036697248e-05,
      "loss": 10.3251,
      "step": 39
    },
    {
      "epoch": 1.0125786163522013,
      "grad_norm": 0.4300397038459778,
      "learning_rate": 7.320183486238533e-05,
      "loss": 16.2943,
      "step": 40
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 0.19085323810577393,
      "learning_rate": 7.227522935779817e-05,
      "loss": 9.4535,
      "step": 41
    },
    {
      "epoch": 1.0628930817610063,
      "grad_norm": 0.1255641132593155,
      "learning_rate": 7.134862385321102e-05,
      "loss": 10.7832,
      "step": 42
    },
    {
      "epoch": 1.0880503144654088,
      "grad_norm": 0.1516360491514206,
      "learning_rate": 7.042201834862386e-05,
      "loss": 10.5093,
      "step": 43
    },
    {
      "epoch": 1.1132075471698113,
      "grad_norm": 0.2405787855386734,
      "learning_rate": 6.949541284403669e-05,
      "loss": 10.0203,
      "step": 44
    },
    {
      "epoch": 1.1383647798742138,
      "grad_norm": 0.3532116115093231,
      "learning_rate": 6.856880733944955e-05,
      "loss": 10.5763,
      "step": 45
    },
    {
      "epoch": 1.1635220125786163,
      "grad_norm": 0.2831573188304901,
      "learning_rate": 6.764220183486238e-05,
      "loss": 9.7205,
      "step": 46
    },
    {
      "epoch": 1.1886792452830188,
      "grad_norm": 0.18701517581939697,
      "learning_rate": 6.671559633027524e-05,
      "loss": 10.387,
      "step": 47
    },
    {
      "epoch": 1.2138364779874213,
      "grad_norm": 0.12146440893411636,
      "learning_rate": 6.578899082568807e-05,
      "loss": 10.4843,
      "step": 48
    },
    {
      "epoch": 1.2389937106918238,
      "grad_norm": 0.12593397498130798,
      "learning_rate": 6.486238532110092e-05,
      "loss": 10.5579,
      "step": 49
    },
    {
      "epoch": 1.2641509433962264,
      "grad_norm": 0.1917344033718109,
      "learning_rate": 6.393577981651376e-05,
      "loss": 9.7609,
      "step": 50
    },
    {
      "epoch": 1.2641509433962264,
      "eval_loss": 10.313277244567871,
      "eval_runtime": 0.1754,
      "eval_samples_per_second": 3044.681,
      "eval_steps_per_second": 96.928,
      "step": 50
    },
    {
      "epoch": 1.2893081761006289,
      "grad_norm": 0.34799277782440186,
      "learning_rate": 6.300917431192661e-05,
      "loss": 11.1836,
      "step": 51
    },
    {
      "epoch": 1.3144654088050314,
      "grad_norm": 0.31355059146881104,
      "learning_rate": 6.208256880733945e-05,
      "loss": 6.6889,
      "step": 52
    },
    {
      "epoch": 1.3396226415094339,
      "grad_norm": 0.23540031909942627,
      "learning_rate": 6.11559633027523e-05,
      "loss": 13.2285,
      "step": 53
    },
    {
      "epoch": 1.3647798742138364,
      "grad_norm": 0.14588120579719543,
      "learning_rate": 6.0229357798165135e-05,
      "loss": 10.7637,
      "step": 54
    },
    {
      "epoch": 1.389937106918239,
      "grad_norm": 0.12895844876766205,
      "learning_rate": 5.930275229357799e-05,
      "loss": 10.2831,
      "step": 55
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 0.16842244565486908,
      "learning_rate": 5.8376146788990825e-05,
      "loss": 9.4583,
      "step": 56
    },
    {
      "epoch": 1.440251572327044,
      "grad_norm": 0.380400151014328,
      "learning_rate": 5.744954128440368e-05,
      "loss": 11.541,
      "step": 57
    },
    {
      "epoch": 1.4654088050314464,
      "grad_norm": 0.2583652436733246,
      "learning_rate": 5.6522935779816515e-05,
      "loss": 7.7146,
      "step": 58
    },
    {
      "epoch": 1.490566037735849,
      "grad_norm": 0.19550327956676483,
      "learning_rate": 5.559633027522936e-05,
      "loss": 12.1234,
      "step": 59
    },
    {
      "epoch": 1.5157232704402515,
      "grad_norm": 0.14996886253356934,
      "learning_rate": 5.46697247706422e-05,
      "loss": 10.5247,
      "step": 60
    },
    {
      "epoch": 1.540880503144654,
      "grad_norm": 0.13765446841716766,
      "learning_rate": 5.374311926605505e-05,
      "loss": 10.2347,
      "step": 61
    },
    {
      "epoch": 1.5660377358490565,
      "grad_norm": 0.15372247993946075,
      "learning_rate": 5.281651376146789e-05,
      "loss": 10.319,
      "step": 62
    },
    {
      "epoch": 1.591194968553459,
      "grad_norm": 0.2650212049484253,
      "learning_rate": 5.188990825688074e-05,
      "loss": 10.4112,
      "step": 63
    },
    {
      "epoch": 1.6163522012578615,
      "grad_norm": 0.2876375913619995,
      "learning_rate": 5.096330275229358e-05,
      "loss": 10.5638,
      "step": 64
    },
    {
      "epoch": 1.641509433962264,
      "grad_norm": 0.21278758347034454,
      "learning_rate": 5.0036697247706424e-05,
      "loss": 10.398,
      "step": 65
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.1513729989528656,
      "learning_rate": 4.911009174311927e-05,
      "loss": 9.6802,
      "step": 66
    },
    {
      "epoch": 1.691823899371069,
      "grad_norm": 0.16482418775558472,
      "learning_rate": 4.8183486238532114e-05,
      "loss": 10.7964,
      "step": 67
    },
    {
      "epoch": 1.7169811320754715,
      "grad_norm": 0.16112814843654633,
      "learning_rate": 4.725688073394496e-05,
      "loss": 10.3939,
      "step": 68
    },
    {
      "epoch": 1.742138364779874,
      "grad_norm": 0.175452321767807,
      "learning_rate": 4.6330275229357804e-05,
      "loss": 10.1161,
      "step": 69
    },
    {
      "epoch": 1.7672955974842768,
      "grad_norm": 0.26655417680740356,
      "learning_rate": 4.540366972477065e-05,
      "loss": 10.3799,
      "step": 70
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 0.22874386608600616,
      "learning_rate": 4.447706422018349e-05,
      "loss": 10.1276,
      "step": 71
    },
    {
      "epoch": 1.8176100628930818,
      "grad_norm": 0.18383730947971344,
      "learning_rate": 4.355045871559633e-05,
      "loss": 9.9519,
      "step": 72
    },
    {
      "epoch": 1.8427672955974843,
      "grad_norm": 0.13625063002109528,
      "learning_rate": 4.262385321100918e-05,
      "loss": 10.7582,
      "step": 73
    },
    {
      "epoch": 1.8679245283018868,
      "grad_norm": 0.13668720424175262,
      "learning_rate": 4.169724770642202e-05,
      "loss": 10.263,
      "step": 74
    },
    {
      "epoch": 1.8930817610062893,
      "grad_norm": 0.14149360358715057,
      "learning_rate": 4.077064220183487e-05,
      "loss": 9.9526,
      "step": 75
    },
    {
      "epoch": 1.9182389937106918,
      "grad_norm": 0.31288641691207886,
      "learning_rate": 3.984403669724771e-05,
      "loss": 11.2683,
      "step": 76
    },
    {
      "epoch": 1.9433962264150944,
      "grad_norm": 0.2737663984298706,
      "learning_rate": 3.891743119266055e-05,
      "loss": 6.941,
      "step": 77
    },
    {
      "epoch": 1.9685534591194969,
      "grad_norm": 0.17450223863124847,
      "learning_rate": 3.7990825688073395e-05,
      "loss": 13.6963,
      "step": 78
    },
    {
      "epoch": 1.9937106918238994,
      "grad_norm": 0.256649374961853,
      "learning_rate": 3.706422018348624e-05,
      "loss": 14.5839,
      "step": 79
    },
    {
      "epoch": 2.0251572327044025,
      "grad_norm": 0.1788261979818344,
      "learning_rate": 3.6137614678899085e-05,
      "loss": 10.3272,
      "step": 80
    },
    {
      "epoch": 2.050314465408805,
      "grad_norm": 0.15341153740882874,
      "learning_rate": 3.521100917431193e-05,
      "loss": 10.331,
      "step": 81
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 0.16519047319889069,
      "learning_rate": 3.4284403669724775e-05,
      "loss": 10.3231,
      "step": 82
    },
    {
      "epoch": 2.10062893081761,
      "grad_norm": 0.17606724798679352,
      "learning_rate": 3.335779816513762e-05,
      "loss": 10.3009,
      "step": 83
    },
    {
      "epoch": 2.1257861635220126,
      "grad_norm": 0.29273226857185364,
      "learning_rate": 3.243119266055046e-05,
      "loss": 10.2937,
      "step": 84
    },
    {
      "epoch": 2.150943396226415,
      "grad_norm": 0.33806756138801575,
      "learning_rate": 3.1504587155963303e-05,
      "loss": 10.2861,
      "step": 85
    },
    {
      "epoch": 2.1761006289308176,
      "grad_norm": 0.17116469144821167,
      "learning_rate": 3.057798165137615e-05,
      "loss": 10.322,
      "step": 86
    },
    {
      "epoch": 2.20125786163522,
      "grad_norm": 0.15034303069114685,
      "learning_rate": 2.9651376146788993e-05,
      "loss": 10.3318,
      "step": 87
    },
    {
      "epoch": 2.2264150943396226,
      "grad_norm": 0.15952639281749725,
      "learning_rate": 2.872477064220184e-05,
      "loss": 10.3268,
      "step": 88
    },
    {
      "epoch": 2.251572327044025,
      "grad_norm": 0.16502609848976135,
      "learning_rate": 2.779816513761468e-05,
      "loss": 10.3057,
      "step": 89
    },
    {
      "epoch": 2.2767295597484276,
      "grad_norm": 0.2700464427471161,
      "learning_rate": 2.6871559633027525e-05,
      "loss": 10.2954,
      "step": 90
    },
    {
      "epoch": 2.30188679245283,
      "grad_norm": 0.2575024366378784,
      "learning_rate": 2.594495412844037e-05,
      "loss": 10.2855,
      "step": 91
    },
    {
      "epoch": 2.3270440251572326,
      "grad_norm": 0.15534816682338715,
      "learning_rate": 2.5018348623853212e-05,
      "loss": 10.3186,
      "step": 92
    },
    {
      "epoch": 2.352201257861635,
      "grad_norm": 0.15319371223449707,
      "learning_rate": 2.4091743119266057e-05,
      "loss": 10.3259,
      "step": 93
    },
    {
      "epoch": 2.3773584905660377,
      "grad_norm": 0.13836082816123962,
      "learning_rate": 2.3165137614678902e-05,
      "loss": 10.3322,
      "step": 94
    },
    {
      "epoch": 2.40251572327044,
      "grad_norm": 0.14459159970283508,
      "learning_rate": 2.2238532110091743e-05,
      "loss": 10.3131,
      "step": 95
    },
    {
      "epoch": 2.4276729559748427,
      "grad_norm": 0.15748228132724762,
      "learning_rate": 2.131192660550459e-05,
      "loss": 10.2985,
      "step": 96
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 0.247804194688797,
      "learning_rate": 2.0385321100917433e-05,
      "loss": 10.2892,
      "step": 97
    },
    {
      "epoch": 2.4779874213836477,
      "grad_norm": 0.17554815113544464,
      "learning_rate": 1.9458715596330275e-05,
      "loss": 10.3041,
      "step": 98
    },
    {
      "epoch": 2.50314465408805,
      "grad_norm": 0.1803831309080124,
      "learning_rate": 1.853211009174312e-05,
      "loss": 10.3223,
      "step": 99
    },
    {
      "epoch": 2.5283018867924527,
      "grad_norm": 0.1464659571647644,
      "learning_rate": 1.7605504587155965e-05,
      "loss": 10.3293,
      "step": 100
    },
    {
      "epoch": 2.5283018867924527,
      "eval_loss": 10.305556297302246,
      "eval_runtime": 0.1774,
      "eval_samples_per_second": 3010.845,
      "eval_steps_per_second": 95.851,
      "step": 100
    },
    {
      "epoch": 2.5534591194968552,
      "grad_norm": 0.1466299146413803,
      "learning_rate": 1.667889908256881e-05,
      "loss": 10.3158,
      "step": 101
    },
    {
      "epoch": 2.5786163522012577,
      "grad_norm": 0.17850784957408905,
      "learning_rate": 1.5752293577981652e-05,
      "loss": 10.2993,
      "step": 102
    },
    {
      "epoch": 2.6037735849056602,
      "grad_norm": 0.2629879117012024,
      "learning_rate": 1.4825688073394497e-05,
      "loss": 10.2901,
      "step": 103
    },
    {
      "epoch": 2.6289308176100628,
      "grad_norm": 0.3722081184387207,
      "learning_rate": 1.389908256880734e-05,
      "loss": 10.2754,
      "step": 104
    },
    {
      "epoch": 2.6540880503144653,
      "grad_norm": 0.17005372047424316,
      "learning_rate": 1.2972477064220185e-05,
      "loss": 10.3255,
      "step": 105
    },
    {
      "epoch": 2.6792452830188678,
      "grad_norm": 0.14880169928073883,
      "learning_rate": 1.2045871559633028e-05,
      "loss": 10.3339,
      "step": 106
    },
    {
      "epoch": 2.7044025157232703,
      "grad_norm": 0.18930739164352417,
      "learning_rate": 1.1119266055045872e-05,
      "loss": 10.3178,
      "step": 107
    },
    {
      "epoch": 2.729559748427673,
      "grad_norm": 0.18744437396526337,
      "learning_rate": 1.0192660550458717e-05,
      "loss": 10.2987,
      "step": 108
    },
    {
      "epoch": 2.7547169811320753,
      "grad_norm": 0.27432578802108765,
      "learning_rate": 9.26605504587156e-06,
      "loss": 10.2927,
      "step": 109
    },
    {
      "epoch": 2.779874213836478,
      "grad_norm": 0.31490325927734375,
      "learning_rate": 8.339449541284405e-06,
      "loss": 10.2855,
      "step": 110
    },
    {
      "epoch": 2.8050314465408803,
      "grad_norm": 0.17064593732357025,
      "learning_rate": 7.412844036697248e-06,
      "loss": 10.3155,
      "step": 111
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 0.15425540506839752,
      "learning_rate": 6.4862385321100925e-06,
      "loss": 10.3282,
      "step": 112
    },
    {
      "epoch": 2.8553459119496853,
      "grad_norm": 0.14992254972457886,
      "learning_rate": 5.559633027522936e-06,
      "loss": 10.3263,
      "step": 113
    },
    {
      "epoch": 2.880503144654088,
      "grad_norm": 0.1558142602443695,
      "learning_rate": 4.63302752293578e-06,
      "loss": 10.3093,
      "step": 114
    },
    {
      "epoch": 2.9056603773584904,
      "grad_norm": 0.2051534205675125,
      "learning_rate": 3.706422018348624e-06,
      "loss": 10.2948,
      "step": 115
    },
    {
      "epoch": 2.930817610062893,
      "grad_norm": 0.2544957101345062,
      "learning_rate": 2.779816513761468e-06,
      "loss": 10.2848,
      "step": 116
    },
    {
      "epoch": 2.9559748427672954,
      "grad_norm": 0.13796408474445343,
      "learning_rate": 1.853211009174312e-06,
      "loss": 10.3145,
      "step": 117
    },
    {
      "epoch": 2.981132075471698,
      "grad_norm": 0.12653085589408875,
      "learning_rate": 9.26605504587156e-07,
      "loss": 10.3105,
      "step": 118
    },
    {
      "epoch": 3.0125786163522013,
      "grad_norm": 0.185125932097435,
      "learning_rate": 0.0,
      "loss": 16.0677,
      "step": 119
    }
  ],
  "logging_steps": 1,
  "max_steps": 119,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 206642570330112.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}