{
  "best_metric": 2.0331459045410156,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.27605244996549344,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013802622498274672,
      "grad_norm": 1.3106821775436401,
      "learning_rate": 1.004e-05,
      "loss": 3.6649,
      "step": 1
    },
    {
      "epoch": 0.0013802622498274672,
      "eval_loss": 3.9833779335021973,
      "eval_runtime": 7.1368,
      "eval_samples_per_second": 42.736,
      "eval_steps_per_second": 10.789,
      "step": 1
    },
    {
      "epoch": 0.0027605244996549345,
      "grad_norm": 1.2568550109863281,
      "learning_rate": 2.008e-05,
      "loss": 3.4178,
      "step": 2
    },
    {
      "epoch": 0.004140786749482402,
      "grad_norm": 1.7689608335494995,
      "learning_rate": 3.012e-05,
      "loss": 4.1951,
      "step": 3
    },
    {
      "epoch": 0.005521048999309869,
      "grad_norm": 1.5144968032836914,
      "learning_rate": 4.016e-05,
      "loss": 3.4457,
      "step": 4
    },
    {
      "epoch": 0.006901311249137336,
      "grad_norm": 1.4309552907943726,
      "learning_rate": 5.02e-05,
      "loss": 3.1559,
      "step": 5
    },
    {
      "epoch": 0.008281573498964804,
      "grad_norm": 1.4668313264846802,
      "learning_rate": 6.024e-05,
      "loss": 3.1952,
      "step": 6
    },
    {
      "epoch": 0.00966183574879227,
      "grad_norm": 1.4052668809890747,
      "learning_rate": 7.028e-05,
      "loss": 3.7713,
      "step": 7
    },
    {
      "epoch": 0.011042097998619738,
      "grad_norm": 1.7720158100128174,
      "learning_rate": 8.032e-05,
      "loss": 3.3666,
      "step": 8
    },
    {
      "epoch": 0.012422360248447204,
      "grad_norm": 1.3920046091079712,
      "learning_rate": 9.036000000000001e-05,
      "loss": 3.0858,
      "step": 9
    },
    {
      "epoch": 0.013802622498274672,
      "grad_norm": 1.6270596981048584,
      "learning_rate": 0.0001004,
      "loss": 3.4722,
      "step": 10
    },
    {
      "epoch": 0.01518288474810214,
      "grad_norm": 1.4511244297027588,
      "learning_rate": 9.987157894736842e-05,
      "loss": 3.1096,
      "step": 11
    },
    {
      "epoch": 0.016563146997929608,
      "grad_norm": 1.6646735668182373,
      "learning_rate": 9.934315789473684e-05,
      "loss": 3.1367,
      "step": 12
    },
    {
      "epoch": 0.017943409247757072,
      "grad_norm": 1.7026424407958984,
      "learning_rate": 9.881473684210525e-05,
      "loss": 3.3196,
      "step": 13
    },
    {
      "epoch": 0.01932367149758454,
      "grad_norm": 1.546176791191101,
      "learning_rate": 9.828631578947369e-05,
      "loss": 2.9841,
      "step": 14
    },
    {
      "epoch": 0.020703933747412008,
      "grad_norm": 1.343595027923584,
      "learning_rate": 9.77578947368421e-05,
      "loss": 3.0285,
      "step": 15
    },
    {
      "epoch": 0.022084195997239476,
      "grad_norm": 1.4044378995895386,
      "learning_rate": 9.722947368421052e-05,
      "loss": 2.7441,
      "step": 16
    },
    {
      "epoch": 0.023464458247066944,
      "grad_norm": 1.4049690961837769,
      "learning_rate": 9.670105263157895e-05,
      "loss": 3.085,
      "step": 17
    },
    {
      "epoch": 0.024844720496894408,
      "grad_norm": 1.8391739130020142,
      "learning_rate": 9.617263157894737e-05,
      "loss": 2.3826,
      "step": 18
    },
    {
      "epoch": 0.026224982746721876,
      "grad_norm": 1.7621740102767944,
      "learning_rate": 9.564421052631579e-05,
      "loss": 3.0509,
      "step": 19
    },
    {
      "epoch": 0.027605244996549344,
      "grad_norm": 1.8850064277648926,
      "learning_rate": 9.511578947368421e-05,
      "loss": 2.6851,
      "step": 20
    },
    {
      "epoch": 0.028985507246376812,
      "grad_norm": 2.0599141120910645,
      "learning_rate": 9.458736842105264e-05,
      "loss": 2.9651,
      "step": 21
    },
    {
      "epoch": 0.03036576949620428,
      "grad_norm": 1.4605776071548462,
      "learning_rate": 9.405894736842106e-05,
      "loss": 3.0117,
      "step": 22
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 1.492771863937378,
      "learning_rate": 9.353052631578947e-05,
      "loss": 2.2637,
      "step": 23
    },
    {
      "epoch": 0.033126293995859216,
      "grad_norm": 1.553098201751709,
      "learning_rate": 9.300210526315789e-05,
      "loss": 2.8497,
      "step": 24
    },
    {
      "epoch": 0.03450655624568668,
      "grad_norm": 1.5250822305679321,
      "learning_rate": 9.247368421052631e-05,
      "loss": 2.5029,
      "step": 25
    },
    {
      "epoch": 0.035886818495514144,
      "grad_norm": 1.6337836980819702,
      "learning_rate": 9.194526315789473e-05,
      "loss": 2.6613,
      "step": 26
    },
    {
      "epoch": 0.037267080745341616,
      "grad_norm": 1.2553786039352417,
      "learning_rate": 9.141684210526316e-05,
      "loss": 2.7623,
      "step": 27
    },
    {
      "epoch": 0.03864734299516908,
      "grad_norm": 1.3037574291229248,
      "learning_rate": 9.088842105263158e-05,
      "loss": 2.6177,
      "step": 28
    },
    {
      "epoch": 0.04002760524499655,
      "grad_norm": 1.4453455209732056,
      "learning_rate": 9.036000000000001e-05,
      "loss": 2.7302,
      "step": 29
    },
    {
      "epoch": 0.041407867494824016,
      "grad_norm": 1.390273928642273,
      "learning_rate": 8.983157894736843e-05,
      "loss": 2.734,
      "step": 30
    },
    {
      "epoch": 0.04278812974465148,
      "grad_norm": 1.2343300580978394,
      "learning_rate": 8.930315789473684e-05,
      "loss": 2.6957,
      "step": 31
    },
    {
      "epoch": 0.04416839199447895,
      "grad_norm": 1.5033609867095947,
      "learning_rate": 8.877473684210526e-05,
      "loss": 2.7113,
      "step": 32
    },
    {
      "epoch": 0.045548654244306416,
      "grad_norm": 1.31312096118927,
      "learning_rate": 8.824631578947368e-05,
      "loss": 2.2402,
      "step": 33
    },
    {
      "epoch": 0.04692891649413389,
      "grad_norm": 1.6782886981964111,
      "learning_rate": 8.771789473684211e-05,
      "loss": 3.0312,
      "step": 34
    },
    {
      "epoch": 0.04830917874396135,
      "grad_norm": 1.358025074005127,
      "learning_rate": 8.718947368421053e-05,
      "loss": 2.9553,
      "step": 35
    },
    {
      "epoch": 0.049689440993788817,
      "grad_norm": 1.3450242280960083,
      "learning_rate": 8.666105263157895e-05,
      "loss": 2.1944,
      "step": 36
    },
    {
      "epoch": 0.05106970324361629,
      "grad_norm": 1.7064541578292847,
      "learning_rate": 8.613263157894737e-05,
      "loss": 2.8165,
      "step": 37
    },
    {
      "epoch": 0.05244996549344375,
      "grad_norm": 1.2318989038467407,
      "learning_rate": 8.560421052631578e-05,
      "loss": 2.3564,
      "step": 38
    },
    {
      "epoch": 0.053830227743271224,
      "grad_norm": 1.4910681247711182,
      "learning_rate": 8.50757894736842e-05,
      "loss": 2.3236,
      "step": 39
    },
    {
      "epoch": 0.05521048999309869,
      "grad_norm": 1.3855364322662354,
      "learning_rate": 8.454736842105263e-05,
      "loss": 2.0899,
      "step": 40
    },
    {
      "epoch": 0.05659075224292615,
      "grad_norm": 1.4245091676712036,
      "learning_rate": 8.401894736842106e-05,
      "loss": 2.5376,
      "step": 41
    },
    {
      "epoch": 0.057971014492753624,
      "grad_norm": 1.459370732307434,
      "learning_rate": 8.349052631578948e-05,
      "loss": 2.4498,
      "step": 42
    },
    {
      "epoch": 0.05935127674258109,
      "grad_norm": 1.5853793621063232,
      "learning_rate": 8.29621052631579e-05,
      "loss": 2.2517,
      "step": 43
    },
    {
      "epoch": 0.06073153899240856,
      "grad_norm": 1.3990416526794434,
      "learning_rate": 8.243368421052632e-05,
      "loss": 2.2779,
      "step": 44
    },
    {
      "epoch": 0.062111801242236024,
      "grad_norm": 1.500887155532837,
      "learning_rate": 8.190526315789474e-05,
      "loss": 2.6741,
      "step": 45
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 2.369209051132202,
      "learning_rate": 8.137684210526315e-05,
      "loss": 2.6675,
      "step": 46
    },
    {
      "epoch": 0.06487232574189096,
      "grad_norm": 1.673604965209961,
      "learning_rate": 8.084842105263157e-05,
      "loss": 1.6952,
      "step": 47
    },
    {
      "epoch": 0.06625258799171843,
      "grad_norm": 2.053783655166626,
      "learning_rate": 8.032e-05,
      "loss": 1.8816,
      "step": 48
    },
    {
      "epoch": 0.06763285024154589,
      "grad_norm": 2.251725435256958,
      "learning_rate": 7.979157894736842e-05,
      "loss": 2.3692,
      "step": 49
    },
    {
      "epoch": 0.06901311249137336,
      "grad_norm": 3.693898916244507,
      "learning_rate": 7.926315789473684e-05,
      "loss": 2.2844,
      "step": 50
    },
    {
      "epoch": 0.06901311249137336,
      "eval_loss": 2.2726948261260986,
      "eval_runtime": 7.0598,
      "eval_samples_per_second": 43.202,
      "eval_steps_per_second": 10.907,
      "step": 50
    },
    {
      "epoch": 0.07039337474120083,
      "grad_norm": 2.035337209701538,
      "learning_rate": 7.873473684210526e-05,
      "loss": 2.4293,
      "step": 51
    },
    {
      "epoch": 0.07177363699102829,
      "grad_norm": 1.4800270795822144,
      "learning_rate": 7.820631578947369e-05,
      "loss": 1.8436,
      "step": 52
    },
    {
      "epoch": 0.07315389924085576,
      "grad_norm": 1.2643930912017822,
      "learning_rate": 7.76778947368421e-05,
      "loss": 2.451,
      "step": 53
    },
    {
      "epoch": 0.07453416149068323,
      "grad_norm": 1.1129631996154785,
      "learning_rate": 7.714947368421052e-05,
      "loss": 2.1686,
      "step": 54
    },
    {
      "epoch": 0.0759144237405107,
      "grad_norm": 1.9466997385025024,
      "learning_rate": 7.662105263157896e-05,
      "loss": 3.0566,
      "step": 55
    },
    {
      "epoch": 0.07729468599033816,
      "grad_norm": 1.585560917854309,
      "learning_rate": 7.609263157894737e-05,
      "loss": 2.9491,
      "step": 56
    },
    {
      "epoch": 0.07867494824016563,
      "grad_norm": 1.148253321647644,
      "learning_rate": 7.556421052631579e-05,
      "loss": 2.5166,
      "step": 57
    },
    {
      "epoch": 0.0800552104899931,
      "grad_norm": 1.3044848442077637,
      "learning_rate": 7.503578947368421e-05,
      "loss": 2.061,
      "step": 58
    },
    {
      "epoch": 0.08143547273982056,
      "grad_norm": 1.3018648624420166,
      "learning_rate": 7.450736842105263e-05,
      "loss": 2.2174,
      "step": 59
    },
    {
      "epoch": 0.08281573498964803,
      "grad_norm": 1.179358720779419,
      "learning_rate": 7.397894736842105e-05,
      "loss": 1.8668,
      "step": 60
    },
    {
      "epoch": 0.0841959972394755,
      "grad_norm": 1.4359699487686157,
      "learning_rate": 7.345052631578948e-05,
      "loss": 1.8076,
      "step": 61
    },
    {
      "epoch": 0.08557625948930296,
      "grad_norm": 1.429067611694336,
      "learning_rate": 7.29221052631579e-05,
      "loss": 2.0383,
      "step": 62
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 1.4482382535934448,
      "learning_rate": 7.239368421052631e-05,
      "loss": 2.5015,
      "step": 63
    },
    {
      "epoch": 0.0883367839889579,
      "grad_norm": 1.7771934270858765,
      "learning_rate": 7.186526315789474e-05,
      "loss": 2.2055,
      "step": 64
    },
    {
      "epoch": 0.08971704623878538,
      "grad_norm": 1.0457968711853027,
      "learning_rate": 7.133684210526316e-05,
      "loss": 1.5984,
      "step": 65
    },
    {
      "epoch": 0.09109730848861283,
      "grad_norm": 1.2737748622894287,
      "learning_rate": 7.080842105263158e-05,
      "loss": 1.7775,
      "step": 66
    },
    {
      "epoch": 0.0924775707384403,
      "grad_norm": 1.1890259981155396,
      "learning_rate": 7.028e-05,
      "loss": 2.0228,
      "step": 67
    },
    {
      "epoch": 0.09385783298826778,
      "grad_norm": 1.2597479820251465,
      "learning_rate": 6.975157894736843e-05,
      "loss": 2.0629,
      "step": 68
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 1.0471433401107788,
      "learning_rate": 6.922315789473685e-05,
      "loss": 1.7882,
      "step": 69
    },
    {
      "epoch": 0.0966183574879227,
      "grad_norm": 1.0744673013687134,
      "learning_rate": 6.869473684210527e-05,
      "loss": 1.7844,
      "step": 70
    },
    {
      "epoch": 0.09799861973775018,
      "grad_norm": 1.393317699432373,
      "learning_rate": 6.816631578947368e-05,
      "loss": 2.1979,
      "step": 71
    },
    {
      "epoch": 0.09937888198757763,
      "grad_norm": 1.4898236989974976,
      "learning_rate": 6.76378947368421e-05,
      "loss": 2.5061,
      "step": 72
    },
    {
      "epoch": 0.1007591442374051,
      "grad_norm": 1.1046738624572754,
      "learning_rate": 6.710947368421052e-05,
      "loss": 1.8938,
      "step": 73
    },
    {
      "epoch": 0.10213940648723258,
      "grad_norm": 1.3718451261520386,
      "learning_rate": 6.658105263157894e-05,
      "loss": 2.5167,
      "step": 74
    },
    {
      "epoch": 0.10351966873706005,
      "grad_norm": 1.2800939083099365,
      "learning_rate": 6.605263157894737e-05,
      "loss": 1.8427,
      "step": 75
    },
    {
      "epoch": 0.1048999309868875,
      "grad_norm": 1.2824013233184814,
      "learning_rate": 6.55242105263158e-05,
      "loss": 1.6855,
      "step": 76
    },
    {
      "epoch": 0.10628019323671498,
      "grad_norm": 1.1829575300216675,
      "learning_rate": 6.499578947368422e-05,
      "loss": 2.1127,
      "step": 77
    },
    {
      "epoch": 0.10766045548654245,
      "grad_norm": 1.4018757343292236,
      "learning_rate": 6.446736842105264e-05,
      "loss": 2.6736,
      "step": 78
    },
    {
      "epoch": 0.1090407177363699,
      "grad_norm": 1.1487196683883667,
      "learning_rate": 6.393894736842105e-05,
      "loss": 2.1348,
      "step": 79
    },
    {
      "epoch": 0.11042097998619738,
      "grad_norm": 1.2489293813705444,
      "learning_rate": 6.341052631578947e-05,
      "loss": 2.4025,
      "step": 80
    },
    {
      "epoch": 0.11180124223602485,
      "grad_norm": 1.379892349243164,
      "learning_rate": 6.288210526315789e-05,
      "loss": 2.1926,
      "step": 81
    },
    {
      "epoch": 0.1131815044858523,
      "grad_norm": 1.3127739429473877,
      "learning_rate": 6.235368421052632e-05,
      "loss": 1.6522,
      "step": 82
    },
    {
      "epoch": 0.11456176673567978,
      "grad_norm": 1.0042823553085327,
      "learning_rate": 6.182526315789474e-05,
      "loss": 1.452,
      "step": 83
    },
    {
      "epoch": 0.11594202898550725,
      "grad_norm": 1.0960407257080078,
      "learning_rate": 6.129684210526316e-05,
      "loss": 1.8077,
      "step": 84
    },
    {
      "epoch": 0.11732229123533472,
      "grad_norm": 1.3931646347045898,
      "learning_rate": 6.076842105263158e-05,
      "loss": 1.9234,
      "step": 85
    },
    {
      "epoch": 0.11870255348516218,
      "grad_norm": 1.2463769912719727,
      "learning_rate": 6.024e-05,
      "loss": 2.3322,
      "step": 86
    },
    {
      "epoch": 0.12008281573498965,
      "grad_norm": 1.5501362085342407,
      "learning_rate": 5.971157894736842e-05,
      "loss": 2.214,
      "step": 87
    },
    {
      "epoch": 0.12146307798481712,
      "grad_norm": 1.4971249103546143,
      "learning_rate": 5.9183157894736835e-05,
      "loss": 2.1229,
      "step": 88
    },
    {
      "epoch": 0.12284334023464458,
      "grad_norm": 1.2540746927261353,
      "learning_rate": 5.8654736842105267e-05,
      "loss": 2.319,
      "step": 89
    },
    {
      "epoch": 0.12422360248447205,
      "grad_norm": 1.3296970129013062,
      "learning_rate": 5.8126315789473684e-05,
      "loss": 2.2295,
      "step": 90
    },
    {
      "epoch": 0.12560386473429952,
      "grad_norm": 1.635713815689087,
      "learning_rate": 5.759789473684211e-05,
      "loss": 2.0453,
      "step": 91
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 2.5068306922912598,
      "learning_rate": 5.706947368421053e-05,
      "loss": 3.5522,
      "step": 92
    },
    {
      "epoch": 0.12836438923395446,
      "grad_norm": 1.3130208253860474,
      "learning_rate": 5.6541052631578945e-05,
      "loss": 1.7299,
      "step": 93
    },
    {
      "epoch": 0.12974465148378192,
      "grad_norm": 1.49214768409729,
      "learning_rate": 5.601263157894736e-05,
      "loss": 1.9252,
      "step": 94
    },
    {
      "epoch": 0.13112491373360938,
      "grad_norm": 1.5757135152816772,
      "learning_rate": 5.5484210526315794e-05,
      "loss": 2.0456,
      "step": 95
    },
    {
      "epoch": 0.13250517598343686,
      "grad_norm": 1.3604228496551514,
      "learning_rate": 5.495578947368421e-05,
      "loss": 1.2423,
      "step": 96
    },
    {
      "epoch": 0.13388543823326432,
      "grad_norm": 1.8302099704742432,
      "learning_rate": 5.442736842105264e-05,
      "loss": 2.4817,
      "step": 97
    },
    {
      "epoch": 0.13526570048309178,
      "grad_norm": 1.6365114450454712,
      "learning_rate": 5.3898947368421055e-05,
      "loss": 2.0664,
      "step": 98
    },
    {
      "epoch": 0.13664596273291926,
      "grad_norm": 1.6933999061584473,
      "learning_rate": 5.337052631578947e-05,
      "loss": 1.6216,
      "step": 99
    },
    {
      "epoch": 0.13802622498274672,
      "grad_norm": 3.5092623233795166,
      "learning_rate": 5.284210526315789e-05,
      "loss": 2.9275,
      "step": 100
    },
    {
      "epoch": 0.13802622498274672,
      "eval_loss": 2.09364652633667,
      "eval_runtime": 7.1227,
      "eval_samples_per_second": 42.821,
      "eval_steps_per_second": 10.81,
      "step": 100
    },
    {
      "epoch": 0.13940648723257418,
      "grad_norm": 1.1113362312316895,
      "learning_rate": 5.231368421052631e-05,
      "loss": 1.6735,
      "step": 101
    },
    {
      "epoch": 0.14078674948240166,
      "grad_norm": 1.1717188358306885,
      "learning_rate": 5.178526315789474e-05,
      "loss": 1.8926,
      "step": 102
    },
    {
      "epoch": 0.14216701173222912,
      "grad_norm": 1.1026091575622559,
      "learning_rate": 5.1256842105263165e-05,
      "loss": 2.007,
      "step": 103
    },
    {
      "epoch": 0.14354727398205658,
      "grad_norm": 1.0963923931121826,
      "learning_rate": 5.072842105263158e-05,
      "loss": 1.787,
      "step": 104
    },
    {
      "epoch": 0.14492753623188406,
      "grad_norm": 1.5479772090911865,
      "learning_rate": 5.02e-05,
      "loss": 2.1698,
      "step": 105
    },
    {
      "epoch": 0.14630779848171152,
      "grad_norm": 1.0915744304656982,
      "learning_rate": 4.967157894736842e-05,
      "loss": 2.2872,
      "step": 106
    },
    {
      "epoch": 0.147688060731539,
      "grad_norm": 1.1803730726242065,
      "learning_rate": 4.914315789473684e-05,
      "loss": 2.4359,
      "step": 107
    },
    {
      "epoch": 0.14906832298136646,
      "grad_norm": 1.4295563697814941,
      "learning_rate": 4.861473684210526e-05,
      "loss": 2.1041,
      "step": 108
    },
    {
      "epoch": 0.15044858523119392,
      "grad_norm": 1.0557366609573364,
      "learning_rate": 4.8086315789473686e-05,
      "loss": 1.9453,
      "step": 109
    },
    {
      "epoch": 0.1518288474810214,
      "grad_norm": 1.0212247371673584,
      "learning_rate": 4.7557894736842104e-05,
      "loss": 1.8541,
      "step": 110
    },
    {
      "epoch": 0.15320910973084886,
      "grad_norm": 1.0571060180664062,
      "learning_rate": 4.702947368421053e-05,
      "loss": 1.9645,
      "step": 111
    },
    {
      "epoch": 0.15458937198067632,
      "grad_norm": 1.12859308719635,
      "learning_rate": 4.6501052631578946e-05,
      "loss": 1.8444,
      "step": 112
    },
    {
      "epoch": 0.1559696342305038,
      "grad_norm": 1.2056413888931274,
      "learning_rate": 4.5972631578947364e-05,
      "loss": 1.7649,
      "step": 113
    },
    {
      "epoch": 0.15734989648033126,
      "grad_norm": 1.6534359455108643,
      "learning_rate": 4.544421052631579e-05,
      "loss": 2.2768,
      "step": 114
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 1.90774667263031,
      "learning_rate": 4.4915789473684213e-05,
      "loss": 2.1505,
      "step": 115
    },
    {
      "epoch": 0.1601104209799862,
      "grad_norm": 0.938823401927948,
      "learning_rate": 4.438736842105263e-05,
      "loss": 1.8724,
      "step": 116
    },
    {
      "epoch": 0.16149068322981366,
      "grad_norm": 1.4365911483764648,
      "learning_rate": 4.3858947368421056e-05,
      "loss": 1.879,
      "step": 117
    },
    {
      "epoch": 0.16287094547964112,
      "grad_norm": 1.195896863937378,
      "learning_rate": 4.3330526315789474e-05,
      "loss": 1.9316,
      "step": 118
    },
    {
      "epoch": 0.1642512077294686,
      "grad_norm": 1.1968637704849243,
      "learning_rate": 4.280210526315789e-05,
      "loss": 1.9372,
      "step": 119
    },
    {
      "epoch": 0.16563146997929606,
      "grad_norm": 1.14974045753479,
      "learning_rate": 4.2273684210526317e-05,
      "loss": 1.7375,
      "step": 120
    },
    {
      "epoch": 0.16701173222912352,
      "grad_norm": 1.0331716537475586,
      "learning_rate": 4.174526315789474e-05,
      "loss": 1.4381,
      "step": 121
    },
    {
      "epoch": 0.168391994478951,
      "grad_norm": 1.178269624710083,
      "learning_rate": 4.121684210526316e-05,
      "loss": 1.7424,
      "step": 122
    },
    {
      "epoch": 0.16977225672877846,
      "grad_norm": 1.0635954141616821,
      "learning_rate": 4.068842105263158e-05,
      "loss": 1.8893,
      "step": 123
    },
    {
      "epoch": 0.17115251897860592,
      "grad_norm": 1.6326851844787598,
      "learning_rate": 4.016e-05,
      "loss": 1.811,
      "step": 124
    },
    {
      "epoch": 0.1725327812284334,
      "grad_norm": 1.1465870141983032,
      "learning_rate": 3.963157894736842e-05,
      "loss": 2.0883,
      "step": 125
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 1.2966638803482056,
      "learning_rate": 3.9103157894736844e-05,
      "loss": 2.1553,
      "step": 126
    },
    {
      "epoch": 0.17529330572808835,
      "grad_norm": 1.4659764766693115,
      "learning_rate": 3.857473684210526e-05,
      "loss": 1.6968,
      "step": 127
    },
    {
      "epoch": 0.1766735679779158,
      "grad_norm": 1.5177134275436401,
      "learning_rate": 3.804631578947369e-05,
      "loss": 2.3954,
      "step": 128
    },
    {
      "epoch": 0.17805383022774326,
      "grad_norm": 1.3623806238174438,
      "learning_rate": 3.7517894736842105e-05,
      "loss": 1.9746,
      "step": 129
    },
    {
      "epoch": 0.17943409247757075,
      "grad_norm": 2.719251871109009,
      "learning_rate": 3.698947368421052e-05,
      "loss": 2.2148,
      "step": 130
    },
    {
      "epoch": 0.1808143547273982,
      "grad_norm": 1.3998181819915771,
      "learning_rate": 3.646105263157895e-05,
      "loss": 1.9759,
      "step": 131
    },
    {
      "epoch": 0.18219461697722567,
      "grad_norm": 1.0175858736038208,
      "learning_rate": 3.593263157894737e-05,
      "loss": 1.7491,
      "step": 132
    },
    {
      "epoch": 0.18357487922705315,
      "grad_norm": 1.354689121246338,
      "learning_rate": 3.540421052631579e-05,
      "loss": 2.3992,
      "step": 133
    },
    {
      "epoch": 0.1849551414768806,
      "grad_norm": 1.246686577796936,
      "learning_rate": 3.4875789473684215e-05,
      "loss": 1.6956,
      "step": 134
    },
    {
      "epoch": 0.18633540372670807,
      "grad_norm": 1.4282056093215942,
      "learning_rate": 3.434736842105263e-05,
      "loss": 1.8083,
      "step": 135
    },
    {
      "epoch": 0.18771566597653555,
      "grad_norm": 1.2194526195526123,
      "learning_rate": 3.381894736842105e-05,
      "loss": 1.5603,
      "step": 136
    },
    {
      "epoch": 0.189095928226363,
      "grad_norm": 1.3371927738189697,
      "learning_rate": 3.329052631578947e-05,
      "loss": 1.8759,
      "step": 137
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 1.5761876106262207,
      "learning_rate": 3.27621052631579e-05,
      "loss": 1.8185,
      "step": 138
    },
    {
      "epoch": 0.19185645272601795,
      "grad_norm": 1.40762197971344,
      "learning_rate": 3.223368421052632e-05,
      "loss": 2.1374,
      "step": 139
    },
    {
      "epoch": 0.1932367149758454,
      "grad_norm": 1.4383774995803833,
      "learning_rate": 3.1705263157894736e-05,
      "loss": 2.0446,
      "step": 140
    },
    {
      "epoch": 0.19461697722567287,
      "grad_norm": 1.4140267372131348,
      "learning_rate": 3.117684210526316e-05,
      "loss": 2.1209,
      "step": 141
    },
    {
      "epoch": 0.19599723947550035,
      "grad_norm": 1.440808892250061,
      "learning_rate": 3.064842105263158e-05,
      "loss": 1.8894,
      "step": 142
    },
    {
      "epoch": 0.1973775017253278,
      "grad_norm": 1.4002034664154053,
      "learning_rate": 3.012e-05,
      "loss": 1.5099,
      "step": 143
    },
    {
      "epoch": 0.19875776397515527,
      "grad_norm": 1.9405709505081177,
      "learning_rate": 2.9591578947368418e-05,
      "loss": 2.3698,
      "step": 144
    },
    {
      "epoch": 0.20013802622498275,
      "grad_norm": 1.799262285232544,
      "learning_rate": 2.9063157894736842e-05,
      "loss": 2.4062,
      "step": 145
    },
    {
      "epoch": 0.2015182884748102,
      "grad_norm": 1.4573534727096558,
      "learning_rate": 2.8534736842105264e-05,
      "loss": 1.9666,
      "step": 146
    },
    {
      "epoch": 0.2028985507246377,
      "grad_norm": 1.7938553094863892,
      "learning_rate": 2.800631578947368e-05,
      "loss": 1.9005,
      "step": 147
    },
    {
      "epoch": 0.20427881297446515,
      "grad_norm": 1.662430763244629,
      "learning_rate": 2.7477894736842106e-05,
      "loss": 1.8399,
      "step": 148
    },
    {
      "epoch": 0.2056590752242926,
      "grad_norm": 2.1903955936431885,
      "learning_rate": 2.6949473684210527e-05,
      "loss": 2.2387,
      "step": 149
    },
    {
      "epoch": 0.2070393374741201,
      "grad_norm": 2.4292702674865723,
      "learning_rate": 2.6421052631578945e-05,
      "loss": 1.8656,
      "step": 150
    },
    {
      "epoch": 0.2070393374741201,
      "eval_loss": 2.0511386394500732,
      "eval_runtime": 7.5598,
      "eval_samples_per_second": 40.345,
      "eval_steps_per_second": 10.185,
      "step": 150
    },
    {
      "epoch": 0.20841959972394755,
      "grad_norm": 1.4151997566223145,
      "learning_rate": 2.589263157894737e-05,
      "loss": 2.3419,
      "step": 151
    },
    {
      "epoch": 0.209799861973775,
      "grad_norm": 1.3771212100982666,
      "learning_rate": 2.536421052631579e-05,
      "loss": 1.9249,
      "step": 152
    },
    {
      "epoch": 0.2111801242236025,
      "grad_norm": 1.5854957103729248,
      "learning_rate": 2.483578947368421e-05,
      "loss": 2.7915,
      "step": 153
    },
    {
      "epoch": 0.21256038647342995,
      "grad_norm": 1.1987736225128174,
      "learning_rate": 2.430736842105263e-05,
      "loss": 2.0221,
      "step": 154
    },
    {
      "epoch": 0.2139406487232574,
      "grad_norm": 1.370588779449463,
      "learning_rate": 2.3778947368421052e-05,
      "loss": 2.058,
      "step": 155
    },
    {
      "epoch": 0.2153209109730849,
      "grad_norm": 1.2222297191619873,
      "learning_rate": 2.3250526315789473e-05,
      "loss": 1.5635,
      "step": 156
    },
    {
      "epoch": 0.21670117322291235,
      "grad_norm": 1.093675971031189,
      "learning_rate": 2.2722105263157894e-05,
      "loss": 1.9671,
      "step": 157
    },
    {
      "epoch": 0.2180814354727398,
      "grad_norm": 0.9379409551620483,
      "learning_rate": 2.2193684210526316e-05,
      "loss": 1.6508,
      "step": 158
    },
    {
      "epoch": 0.2194616977225673,
      "grad_norm": 1.4368672370910645,
      "learning_rate": 2.1665263157894737e-05,
      "loss": 1.878,
      "step": 159
    },
    {
      "epoch": 0.22084195997239475,
      "grad_norm": 0.921222448348999,
      "learning_rate": 2.1136842105263158e-05,
      "loss": 1.6258,
      "step": 160
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.3184685707092285,
      "learning_rate": 2.060842105263158e-05,
      "loss": 1.8804,
      "step": 161
    },
    {
      "epoch": 0.2236024844720497,
      "grad_norm": 1.3226741552352905,
      "learning_rate": 2.008e-05,
      "loss": 2.0123,
      "step": 162
    },
    {
      "epoch": 0.22498274672187715,
      "grad_norm": 1.1770551204681396,
      "learning_rate": 1.9551578947368422e-05,
      "loss": 2.2692,
      "step": 163
    },
    {
      "epoch": 0.2263630089717046,
      "grad_norm": 1.0083926916122437,
      "learning_rate": 1.9023157894736843e-05,
      "loss": 1.6317,
      "step": 164
    },
    {
      "epoch": 0.2277432712215321,
      "grad_norm": 1.3350708484649658,
      "learning_rate": 1.849473684210526e-05,
      "loss": 1.9381,
      "step": 165
    },
    {
      "epoch": 0.22912353347135955,
      "grad_norm": 1.4198658466339111,
      "learning_rate": 1.7966315789473686e-05,
      "loss": 1.8994,
      "step": 166
    },
    {
      "epoch": 0.23050379572118704,
      "grad_norm": 1.2131656408309937,
      "learning_rate": 1.7437894736842107e-05,
      "loss": 1.6962,
      "step": 167
    },
    {
      "epoch": 0.2318840579710145,
      "grad_norm": 1.5162237882614136,
      "learning_rate": 1.6909473684210525e-05,
      "loss": 2.0727,
      "step": 168
    },
    {
      "epoch": 0.23326432022084195,
      "grad_norm": 0.966874361038208,
      "learning_rate": 1.638105263157895e-05,
      "loss": 1.3924,
      "step": 169
    },
    {
      "epoch": 0.23464458247066944,
      "grad_norm": 1.1200611591339111,
      "learning_rate": 1.5852631578947368e-05,
      "loss": 1.9404,
      "step": 170
    },
    {
      "epoch": 0.2360248447204969,
      "grad_norm": 1.2365270853042603,
      "learning_rate": 1.532421052631579e-05,
      "loss": 2.3918,
      "step": 171
    },
    {
      "epoch": 0.23740510697032435,
      "grad_norm": 1.2294012308120728,
      "learning_rate": 1.4795789473684209e-05,
      "loss": 1.8402,
      "step": 172
    },
    {
      "epoch": 0.23878536922015184,
      "grad_norm": 1.1588753461837769,
      "learning_rate": 1.4267368421052632e-05,
      "loss": 1.3646,
      "step": 173
    },
    {
      "epoch": 0.2401656314699793,
      "grad_norm": 1.1937261819839478,
      "learning_rate": 1.3738947368421053e-05,
      "loss": 1.9111,
      "step": 174
    },
    {
      "epoch": 0.24154589371980675,
      "grad_norm": 1.2816555500030518,
      "learning_rate": 1.3210526315789473e-05,
      "loss": 2.081,
      "step": 175
    },
    {
      "epoch": 0.24292615596963424,
      "grad_norm": 1.1099061965942383,
      "learning_rate": 1.2682105263157896e-05,
      "loss": 1.4477,
      "step": 176
    },
    {
      "epoch": 0.2443064182194617,
      "grad_norm": 1.1234540939331055,
      "learning_rate": 1.2153684210526315e-05,
      "loss": 1.4634,
      "step": 177
    },
    {
      "epoch": 0.24568668046928915,
      "grad_norm": 1.3805010318756104,
      "learning_rate": 1.1625263157894737e-05,
      "loss": 2.0993,
      "step": 178
    },
    {
      "epoch": 0.24706694271911664,
      "grad_norm": 1.942065715789795,
      "learning_rate": 1.1096842105263158e-05,
      "loss": 1.7852,
      "step": 179
    },
    {
      "epoch": 0.2484472049689441,
      "grad_norm": 1.6598256826400757,
      "learning_rate": 1.0568421052631579e-05,
      "loss": 2.3633,
      "step": 180
    },
    {
      "epoch": 0.24982746721877155,
      "grad_norm": 1.2270684242248535,
      "learning_rate": 1.004e-05,
      "loss": 1.7722,
      "step": 181
    },
    {
      "epoch": 0.25120772946859904,
      "grad_norm": 1.3210140466690063,
      "learning_rate": 9.511578947368422e-06,
      "loss": 2.0714,
      "step": 182
    },
    {
      "epoch": 0.2525879917184265,
      "grad_norm": 1.952738881111145,
      "learning_rate": 8.983157894736843e-06,
      "loss": 2.1108,
      "step": 183
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 1.2155840396881104,
      "learning_rate": 8.454736842105263e-06,
      "loss": 2.2641,
      "step": 184
    },
    {
      "epoch": 0.2553485162180814,
      "grad_norm": 1.4390993118286133,
      "learning_rate": 7.926315789473684e-06,
      "loss": 1.9502,
      "step": 185
    },
    {
      "epoch": 0.2567287784679089,
      "grad_norm": 1.2670150995254517,
      "learning_rate": 7.397894736842104e-06,
      "loss": 1.5847,
      "step": 186
    },
    {
      "epoch": 0.2581090407177364,
      "grad_norm": 1.7874329090118408,
      "learning_rate": 6.8694736842105265e-06,
      "loss": 2.0178,
      "step": 187
    },
    {
      "epoch": 0.25948930296756384,
      "grad_norm": 1.404029130935669,
      "learning_rate": 6.341052631578948e-06,
      "loss": 2.1213,
      "step": 188
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 1.2744040489196777,
      "learning_rate": 5.812631578947368e-06,
      "loss": 1.6723,
      "step": 189
    },
    {
      "epoch": 0.26224982746721875,
      "grad_norm": 1.8867708444595337,
      "learning_rate": 5.2842105263157896e-06,
      "loss": 2.7068,
      "step": 190
    },
    {
      "epoch": 0.2636300897170462,
      "grad_norm": 1.2839573621749878,
      "learning_rate": 4.755789473684211e-06,
      "loss": 1.7505,
      "step": 191
    },
    {
      "epoch": 0.2650103519668737,
      "grad_norm": 1.9421333074569702,
      "learning_rate": 4.227368421052631e-06,
      "loss": 2.3668,
      "step": 192
    },
    {
      "epoch": 0.2663906142167012,
      "grad_norm": 1.583764672279358,
      "learning_rate": 3.698947368421052e-06,
      "loss": 1.8675,
      "step": 193
    },
    {
      "epoch": 0.26777087646652864,
      "grad_norm": 1.373460292816162,
      "learning_rate": 3.170526315789474e-06,
      "loss": 1.4618,
      "step": 194
    },
    {
      "epoch": 0.2691511387163561,
      "grad_norm": 1.3068757057189941,
      "learning_rate": 2.6421052631578948e-06,
      "loss": 1.867,
      "step": 195
    },
    {
      "epoch": 0.27053140096618356,
      "grad_norm": 1.478837251663208,
      "learning_rate": 2.1136842105263157e-06,
      "loss": 1.6132,
      "step": 196
    },
    {
      "epoch": 0.27191166321601107,
      "grad_norm": 1.298207402229309,
      "learning_rate": 1.585263157894737e-06,
      "loss": 1.4838,
      "step": 197
    },
    {
      "epoch": 0.2732919254658385,
      "grad_norm": 1.8075675964355469,
      "learning_rate": 1.0568421052631578e-06,
      "loss": 1.6577,
      "step": 198
    },
    {
      "epoch": 0.274672187715666,
      "grad_norm": 1.675578236579895,
      "learning_rate": 5.284210526315789e-07,
      "loss": 1.6484,
      "step": 199
    },
    {
      "epoch": 0.27605244996549344,
      "grad_norm": 2.6218018531799316,
      "learning_rate": 0.0,
      "loss": 2.0445,
      "step": 200
    },
    {
      "epoch": 0.27605244996549344,
      "eval_loss": 2.0331459045410156,
      "eval_runtime": 7.136,
      "eval_samples_per_second": 42.741,
      "eval_steps_per_second": 10.79,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "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": 3434439770112000.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}