{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 123,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008130081300813009,
      "grad_norm": 5.549927234649658,
      "learning_rate": 1e-05,
      "loss": 2.7085,
      "step": 1
    },
    {
      "epoch": 0.016260162601626018,
      "grad_norm": 5.491306781768799,
      "learning_rate": 9.91869918699187e-06,
      "loss": 2.7807,
      "step": 2
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 5.130057334899902,
      "learning_rate": 9.837398373983741e-06,
      "loss": 2.7514,
      "step": 3
    },
    {
      "epoch": 0.032520325203252036,
      "grad_norm": 4.267144680023193,
      "learning_rate": 9.756097560975611e-06,
      "loss": 2.5635,
      "step": 4
    },
    {
      "epoch": 0.04065040650406504,
      "grad_norm": 3.398440361022949,
      "learning_rate": 9.67479674796748e-06,
      "loss": 2.5193,
      "step": 5
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 2.3021163940429688,
      "learning_rate": 9.59349593495935e-06,
      "loss": 2.3388,
      "step": 6
    },
    {
      "epoch": 0.056910569105691054,
      "grad_norm": 1.689666986465454,
      "learning_rate": 9.51219512195122e-06,
      "loss": 2.2621,
      "step": 7
    },
    {
      "epoch": 0.06504065040650407,
      "grad_norm": 1.4065525531768799,
      "learning_rate": 9.43089430894309e-06,
      "loss": 2.3228,
      "step": 8
    },
    {
      "epoch": 0.07317073170731707,
      "grad_norm": 1.2045888900756836,
      "learning_rate": 9.34959349593496e-06,
      "loss": 2.2423,
      "step": 9
    },
    {
      "epoch": 0.08130081300813008,
      "grad_norm": 1.2231402397155762,
      "learning_rate": 9.268292682926831e-06,
      "loss": 2.2951,
      "step": 10
    },
    {
      "epoch": 0.08943089430894309,
      "grad_norm": 1.2492059469223022,
      "learning_rate": 9.1869918699187e-06,
      "loss": 2.1954,
      "step": 11
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 1.2706804275512695,
      "learning_rate": 9.10569105691057e-06,
      "loss": 2.2233,
      "step": 12
    },
    {
      "epoch": 0.10569105691056911,
      "grad_norm": 1.308569073677063,
      "learning_rate": 9.02439024390244e-06,
      "loss": 2.2009,
      "step": 13
    },
    {
      "epoch": 0.11382113821138211,
      "grad_norm": 1.2883176803588867,
      "learning_rate": 8.94308943089431e-06,
      "loss": 2.1515,
      "step": 14
    },
    {
      "epoch": 0.12195121951219512,
      "grad_norm": 1.2327231168746948,
      "learning_rate": 8.86178861788618e-06,
      "loss": 2.0954,
      "step": 15
    },
    {
      "epoch": 0.13008130081300814,
      "grad_norm": 1.1795716285705566,
      "learning_rate": 8.78048780487805e-06,
      "loss": 2.1118,
      "step": 16
    },
    {
      "epoch": 0.13821138211382114,
      "grad_norm": 1.0975069999694824,
      "learning_rate": 8.69918699186992e-06,
      "loss": 2.0454,
      "step": 17
    },
    {
      "epoch": 0.14634146341463414,
      "grad_norm": 1.019369125366211,
      "learning_rate": 8.617886178861789e-06,
      "loss": 2.0432,
      "step": 18
    },
    {
      "epoch": 0.15447154471544716,
      "grad_norm": 0.9480703473091125,
      "learning_rate": 8.536585365853658e-06,
      "loss": 2.0292,
      "step": 19
    },
    {
      "epoch": 0.16260162601626016,
      "grad_norm": 0.9140216708183289,
      "learning_rate": 8.45528455284553e-06,
      "loss": 2.0298,
      "step": 20
    },
    {
      "epoch": 0.17073170731707318,
      "grad_norm": 0.8537278771400452,
      "learning_rate": 8.373983739837399e-06,
      "loss": 2.0035,
      "step": 21
    },
    {
      "epoch": 0.17886178861788618,
      "grad_norm": 0.8466953039169312,
      "learning_rate": 8.292682926829268e-06,
      "loss": 1.949,
      "step": 22
    },
    {
      "epoch": 0.18699186991869918,
      "grad_norm": 0.8279498219490051,
      "learning_rate": 8.21138211382114e-06,
      "loss": 1.948,
      "step": 23
    },
    {
      "epoch": 0.1951219512195122,
      "grad_norm": 0.7655555605888367,
      "learning_rate": 8.130081300813009e-06,
      "loss": 1.916,
      "step": 24
    },
    {
      "epoch": 0.2032520325203252,
      "grad_norm": 0.7242797613143921,
      "learning_rate": 8.048780487804879e-06,
      "loss": 1.9187,
      "step": 25
    },
    {
      "epoch": 0.21138211382113822,
      "grad_norm": 0.6928679943084717,
      "learning_rate": 7.967479674796748e-06,
      "loss": 1.9413,
      "step": 26
    },
    {
      "epoch": 0.21951219512195122,
      "grad_norm": 0.6521593928337097,
      "learning_rate": 7.886178861788618e-06,
      "loss": 1.8127,
      "step": 27
    },
    {
      "epoch": 0.22764227642276422,
      "grad_norm": 0.6483350992202759,
      "learning_rate": 7.804878048780489e-06,
      "loss": 1.8327,
      "step": 28
    },
    {
      "epoch": 0.23577235772357724,
      "grad_norm": 0.6468049883842468,
      "learning_rate": 7.723577235772358e-06,
      "loss": 1.8804,
      "step": 29
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.6291001439094543,
      "learning_rate": 7.64227642276423e-06,
      "loss": 1.8349,
      "step": 30
    },
    {
      "epoch": 0.25203252032520324,
      "grad_norm": 0.6040977835655212,
      "learning_rate": 7.560975609756098e-06,
      "loss": 1.7932,
      "step": 31
    },
    {
      "epoch": 0.2601626016260163,
      "grad_norm": 0.6331721544265747,
      "learning_rate": 7.4796747967479676e-06,
      "loss": 1.85,
      "step": 32
    },
    {
      "epoch": 0.2682926829268293,
      "grad_norm": 0.5773176550865173,
      "learning_rate": 7.398373983739838e-06,
      "loss": 1.7713,
      "step": 33
    },
    {
      "epoch": 0.2764227642276423,
      "grad_norm": 0.5964117050170898,
      "learning_rate": 7.317073170731707e-06,
      "loss": 1.7933,
      "step": 34
    },
    {
      "epoch": 0.2845528455284553,
      "grad_norm": 0.6199694871902466,
      "learning_rate": 7.2357723577235786e-06,
      "loss": 1.8095,
      "step": 35
    },
    {
      "epoch": 0.2926829268292683,
      "grad_norm": 0.601659893989563,
      "learning_rate": 7.154471544715448e-06,
      "loss": 1.8293,
      "step": 36
    },
    {
      "epoch": 0.3008130081300813,
      "grad_norm": 0.5909000635147095,
      "learning_rate": 7.0731707317073175e-06,
      "loss": 1.7735,
      "step": 37
    },
    {
      "epoch": 0.3089430894308943,
      "grad_norm": 0.5953034162521362,
      "learning_rate": 6.991869918699188e-06,
      "loss": 1.7626,
      "step": 38
    },
    {
      "epoch": 0.3170731707317073,
      "grad_norm": 0.5631912350654602,
      "learning_rate": 6.910569105691057e-06,
      "loss": 1.7508,
      "step": 39
    },
    {
      "epoch": 0.3252032520325203,
      "grad_norm": 0.5466800332069397,
      "learning_rate": 6.829268292682928e-06,
      "loss": 1.7447,
      "step": 40
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.5560957193374634,
      "learning_rate": 6.747967479674797e-06,
      "loss": 1.6766,
      "step": 41
    },
    {
      "epoch": 0.34146341463414637,
      "grad_norm": 0.545097827911377,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.6994,
      "step": 42
    },
    {
      "epoch": 0.34959349593495936,
      "grad_norm": 0.5267745852470398,
      "learning_rate": 6.585365853658538e-06,
      "loss": 1.6837,
      "step": 43
    },
    {
      "epoch": 0.35772357723577236,
      "grad_norm": 0.522652804851532,
      "learning_rate": 6.504065040650407e-06,
      "loss": 1.6461,
      "step": 44
    },
    {
      "epoch": 0.36585365853658536,
      "grad_norm": 0.5654343366622925,
      "learning_rate": 6.422764227642278e-06,
      "loss": 1.6596,
      "step": 45
    },
    {
      "epoch": 0.37398373983739835,
      "grad_norm": 0.5501427054405212,
      "learning_rate": 6.341463414634147e-06,
      "loss": 1.6116,
      "step": 46
    },
    {
      "epoch": 0.3821138211382114,
      "grad_norm": 0.483664333820343,
      "learning_rate": 6.260162601626017e-06,
      "loss": 1.6495,
      "step": 47
    },
    {
      "epoch": 0.3902439024390244,
      "grad_norm": 0.48560580611228943,
      "learning_rate": 6.178861788617887e-06,
      "loss": 1.6362,
      "step": 48
    },
    {
      "epoch": 0.3983739837398374,
      "grad_norm": 0.4865823984146118,
      "learning_rate": 6.0975609756097564e-06,
      "loss": 1.6322,
      "step": 49
    },
    {
      "epoch": 0.4065040650406504,
      "grad_norm": 0.4677760601043701,
      "learning_rate": 6.016260162601627e-06,
      "loss": 1.6499,
      "step": 50
    },
    {
      "epoch": 0.4146341463414634,
      "grad_norm": 0.5214702486991882,
      "learning_rate": 5.934959349593496e-06,
      "loss": 1.6878,
      "step": 51
    },
    {
      "epoch": 0.42276422764227645,
      "grad_norm": 0.4658932685852051,
      "learning_rate": 5.853658536585366e-06,
      "loss": 1.6249,
      "step": 52
    },
    {
      "epoch": 0.43089430894308944,
      "grad_norm": 0.4474790096282959,
      "learning_rate": 5.772357723577237e-06,
      "loss": 1.5653,
      "step": 53
    },
    {
      "epoch": 0.43902439024390244,
      "grad_norm": 0.42627304792404175,
      "learning_rate": 5.691056910569106e-06,
      "loss": 1.5974,
      "step": 54
    },
    {
      "epoch": 0.44715447154471544,
      "grad_norm": 0.4498918950557709,
      "learning_rate": 5.609756097560977e-06,
      "loss": 1.5827,
      "step": 55
    },
    {
      "epoch": 0.45528455284552843,
      "grad_norm": 0.42996910214424133,
      "learning_rate": 5.528455284552846e-06,
      "loss": 1.5868,
      "step": 56
    },
    {
      "epoch": 0.4634146341463415,
      "grad_norm": 0.4333815574645996,
      "learning_rate": 5.447154471544716e-06,
      "loss": 1.5496,
      "step": 57
    },
    {
      "epoch": 0.4715447154471545,
      "grad_norm": 0.4117136597633362,
      "learning_rate": 5.365853658536586e-06,
      "loss": 1.5893,
      "step": 58
    },
    {
      "epoch": 0.4796747967479675,
      "grad_norm": 0.39979466795921326,
      "learning_rate": 5.2845528455284555e-06,
      "loss": 1.6002,
      "step": 59
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.40394771099090576,
      "learning_rate": 5.203252032520326e-06,
      "loss": 1.6018,
      "step": 60
    },
    {
      "epoch": 0.4959349593495935,
      "grad_norm": 0.40385180711746216,
      "learning_rate": 5.121951219512195e-06,
      "loss": 1.5461,
      "step": 61
    },
    {
      "epoch": 0.5040650406504065,
      "grad_norm": 0.4215068221092224,
      "learning_rate": 5.040650406504065e-06,
      "loss": 1.5605,
      "step": 62
    },
    {
      "epoch": 0.5121951219512195,
      "grad_norm": 0.37316930294036865,
      "learning_rate": 4.959349593495935e-06,
      "loss": 1.5466,
      "step": 63
    },
    {
      "epoch": 0.5203252032520326,
      "grad_norm": 0.3831072151660919,
      "learning_rate": 4.8780487804878055e-06,
      "loss": 1.5482,
      "step": 64
    },
    {
      "epoch": 0.5284552845528455,
      "grad_norm": 0.4046054482460022,
      "learning_rate": 4.796747967479675e-06,
      "loss": 1.5383,
      "step": 65
    },
    {
      "epoch": 0.5365853658536586,
      "grad_norm": 0.39180001616477966,
      "learning_rate": 4.715447154471545e-06,
      "loss": 1.5113,
      "step": 66
    },
    {
      "epoch": 0.5447154471544715,
      "grad_norm": 0.36748605966567993,
      "learning_rate": 4.634146341463416e-06,
      "loss": 1.5277,
      "step": 67
    },
    {
      "epoch": 0.5528455284552846,
      "grad_norm": 0.39655762910842896,
      "learning_rate": 4.552845528455285e-06,
      "loss": 1.5386,
      "step": 68
    },
    {
      "epoch": 0.5609756097560976,
      "grad_norm": 0.40335220098495483,
      "learning_rate": 4.471544715447155e-06,
      "loss": 1.4758,
      "step": 69
    },
    {
      "epoch": 0.5691056910569106,
      "grad_norm": 0.38374829292297363,
      "learning_rate": 4.390243902439025e-06,
      "loss": 1.5426,
      "step": 70
    },
    {
      "epoch": 0.5772357723577236,
      "grad_norm": 0.387068510055542,
      "learning_rate": 4.308943089430894e-06,
      "loss": 1.5382,
      "step": 71
    },
    {
      "epoch": 0.5853658536585366,
      "grad_norm": 0.395914226770401,
      "learning_rate": 4.227642276422765e-06,
      "loss": 1.4766,
      "step": 72
    },
    {
      "epoch": 0.5934959349593496,
      "grad_norm": 0.3709677755832672,
      "learning_rate": 4.146341463414634e-06,
      "loss": 1.5269,
      "step": 73
    },
    {
      "epoch": 0.6016260162601627,
      "grad_norm": 0.37058818340301514,
      "learning_rate": 4.0650406504065046e-06,
      "loss": 1.5189,
      "step": 74
    },
    {
      "epoch": 0.6097560975609756,
      "grad_norm": 0.36722519993782043,
      "learning_rate": 3.983739837398374e-06,
      "loss": 1.5125,
      "step": 75
    },
    {
      "epoch": 0.6178861788617886,
      "grad_norm": 0.39505788683891296,
      "learning_rate": 3.902439024390244e-06,
      "loss": 1.5597,
      "step": 76
    },
    {
      "epoch": 0.6260162601626016,
      "grad_norm": 0.3756929636001587,
      "learning_rate": 3.821138211382115e-06,
      "loss": 1.4876,
      "step": 77
    },
    {
      "epoch": 0.6341463414634146,
      "grad_norm": 0.363145649433136,
      "learning_rate": 3.7398373983739838e-06,
      "loss": 1.5048,
      "step": 78
    },
    {
      "epoch": 0.6422764227642277,
      "grad_norm": 0.3606300354003906,
      "learning_rate": 3.6585365853658537e-06,
      "loss": 1.5078,
      "step": 79
    },
    {
      "epoch": 0.6504065040650406,
      "grad_norm": 0.3544052541255951,
      "learning_rate": 3.577235772357724e-06,
      "loss": 1.4849,
      "step": 80
    },
    {
      "epoch": 0.6585365853658537,
      "grad_norm": 0.421355664730072,
      "learning_rate": 3.495934959349594e-06,
      "loss": 1.5145,
      "step": 81
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.3610020577907562,
      "learning_rate": 3.414634146341464e-06,
      "loss": 1.466,
      "step": 82
    },
    {
      "epoch": 0.6747967479674797,
      "grad_norm": 0.3549164831638336,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.4849,
      "step": 83
    },
    {
      "epoch": 0.6829268292682927,
      "grad_norm": 0.36329224705696106,
      "learning_rate": 3.2520325203252037e-06,
      "loss": 1.485,
      "step": 84
    },
    {
      "epoch": 0.6910569105691057,
      "grad_norm": 0.35131382942199707,
      "learning_rate": 3.1707317073170736e-06,
      "loss": 1.4716,
      "step": 85
    },
    {
      "epoch": 0.6991869918699187,
      "grad_norm": 0.4444718658924103,
      "learning_rate": 3.0894308943089435e-06,
      "loss": 1.4514,
      "step": 86
    },
    {
      "epoch": 0.7073170731707317,
      "grad_norm": 0.33230820298194885,
      "learning_rate": 3.0081300813008134e-06,
      "loss": 1.4578,
      "step": 87
    },
    {
      "epoch": 0.7154471544715447,
      "grad_norm": 0.37406694889068604,
      "learning_rate": 2.926829268292683e-06,
      "loss": 1.4573,
      "step": 88
    },
    {
      "epoch": 0.7235772357723578,
      "grad_norm": 0.33767250180244446,
      "learning_rate": 2.845528455284553e-06,
      "loss": 1.4489,
      "step": 89
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.3861808776855469,
      "learning_rate": 2.764227642276423e-06,
      "loss": 1.5041,
      "step": 90
    },
    {
      "epoch": 0.7398373983739838,
      "grad_norm": 0.35341161489486694,
      "learning_rate": 2.682926829268293e-06,
      "loss": 1.4769,
      "step": 91
    },
    {
      "epoch": 0.7479674796747967,
      "grad_norm": 0.3487161099910736,
      "learning_rate": 2.601626016260163e-06,
      "loss": 1.4493,
      "step": 92
    },
    {
      "epoch": 0.7560975609756098,
      "grad_norm": 0.34500062465667725,
      "learning_rate": 2.5203252032520324e-06,
      "loss": 1.4079,
      "step": 93
    },
    {
      "epoch": 0.7642276422764228,
      "grad_norm": 0.3244503438472748,
      "learning_rate": 2.4390243902439027e-06,
      "loss": 1.4679,
      "step": 94
    },
    {
      "epoch": 0.7723577235772358,
      "grad_norm": 0.38658860325813293,
      "learning_rate": 2.3577235772357727e-06,
      "loss": 1.494,
      "step": 95
    },
    {
      "epoch": 0.7804878048780488,
      "grad_norm": 0.36310896277427673,
      "learning_rate": 2.2764227642276426e-06,
      "loss": 1.4635,
      "step": 96
    },
    {
      "epoch": 0.7886178861788617,
      "grad_norm": 0.38168150186538696,
      "learning_rate": 2.1951219512195125e-06,
      "loss": 1.4784,
      "step": 97
    },
    {
      "epoch": 0.7967479674796748,
      "grad_norm": 0.3939133584499359,
      "learning_rate": 2.1138211382113824e-06,
      "loss": 1.4702,
      "step": 98
    },
    {
      "epoch": 0.8048780487804879,
      "grad_norm": 0.33597785234451294,
      "learning_rate": 2.0325203252032523e-06,
      "loss": 1.4616,
      "step": 99
    },
    {
      "epoch": 0.8130081300813008,
      "grad_norm": 0.3472687005996704,
      "learning_rate": 1.951219512195122e-06,
      "loss": 1.4346,
      "step": 100
    },
    {
      "epoch": 0.8211382113821138,
      "grad_norm": 0.3384402096271515,
      "learning_rate": 1.8699186991869919e-06,
      "loss": 1.4635,
      "step": 101
    },
    {
      "epoch": 0.8292682926829268,
      "grad_norm": 0.4153139591217041,
      "learning_rate": 1.788617886178862e-06,
      "loss": 1.4916,
      "step": 102
    },
    {
      "epoch": 0.8373983739837398,
      "grad_norm": 0.38573986291885376,
      "learning_rate": 1.707317073170732e-06,
      "loss": 1.4469,
      "step": 103
    },
    {
      "epoch": 0.8455284552845529,
      "grad_norm": 0.3305743634700775,
      "learning_rate": 1.6260162601626018e-06,
      "loss": 1.4691,
      "step": 104
    },
    {
      "epoch": 0.8536585365853658,
      "grad_norm": 0.37890517711639404,
      "learning_rate": 1.5447154471544717e-06,
      "loss": 1.4174,
      "step": 105
    },
    {
      "epoch": 0.8617886178861789,
      "grad_norm": 0.43667566776275635,
      "learning_rate": 1.4634146341463414e-06,
      "loss": 1.5189,
      "step": 106
    },
    {
      "epoch": 0.8699186991869918,
      "grad_norm": 0.3576219975948334,
      "learning_rate": 1.3821138211382116e-06,
      "loss": 1.4618,
      "step": 107
    },
    {
      "epoch": 0.8780487804878049,
      "grad_norm": 0.37747374176979065,
      "learning_rate": 1.3008130081300815e-06,
      "loss": 1.4409,
      "step": 108
    },
    {
      "epoch": 0.8861788617886179,
      "grad_norm": 0.3657553791999817,
      "learning_rate": 1.2195121951219514e-06,
      "loss": 1.4605,
      "step": 109
    },
    {
      "epoch": 0.8943089430894309,
      "grad_norm": 0.33915138244628906,
      "learning_rate": 1.1382113821138213e-06,
      "loss": 1.4353,
      "step": 110
    },
    {
      "epoch": 0.9024390243902439,
      "grad_norm": 0.3774731159210205,
      "learning_rate": 1.0569105691056912e-06,
      "loss": 1.4402,
      "step": 111
    },
    {
      "epoch": 0.9105691056910569,
      "grad_norm": 0.36047080159187317,
      "learning_rate": 9.75609756097561e-07,
      "loss": 1.4587,
      "step": 112
    },
    {
      "epoch": 0.9186991869918699,
      "grad_norm": 0.3694530725479126,
      "learning_rate": 8.94308943089431e-07,
      "loss": 1.4298,
      "step": 113
    },
    {
      "epoch": 0.926829268292683,
      "grad_norm": 0.35272732377052307,
      "learning_rate": 8.130081300813009e-07,
      "loss": 1.4634,
      "step": 114
    },
    {
      "epoch": 0.9349593495934959,
      "grad_norm": 0.36272573471069336,
      "learning_rate": 7.317073170731707e-07,
      "loss": 1.4793,
      "step": 115
    },
    {
      "epoch": 0.943089430894309,
      "grad_norm": 0.3531406819820404,
      "learning_rate": 6.504065040650407e-07,
      "loss": 1.4507,
      "step": 116
    },
    {
      "epoch": 0.9512195121951219,
      "grad_norm": 0.32868847250938416,
      "learning_rate": 5.691056910569106e-07,
      "loss": 1.4514,
      "step": 117
    },
    {
      "epoch": 0.959349593495935,
      "grad_norm": 0.3570950925350189,
      "learning_rate": 4.878048780487805e-07,
      "loss": 1.4402,
      "step": 118
    },
    {
      "epoch": 0.967479674796748,
      "grad_norm": 0.3893052339553833,
      "learning_rate": 4.0650406504065046e-07,
      "loss": 1.4742,
      "step": 119
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.3945332467556,
      "learning_rate": 3.2520325203252037e-07,
      "loss": 1.4649,
      "step": 120
    },
    {
      "epoch": 0.983739837398374,
      "grad_norm": 0.39911824464797974,
      "learning_rate": 2.439024390243903e-07,
      "loss": 1.4045,
      "step": 121
    },
    {
      "epoch": 0.991869918699187,
      "grad_norm": 0.37168797850608826,
      "learning_rate": 1.6260162601626018e-07,
      "loss": 1.4646,
      "step": 122
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34375110268592834,
      "learning_rate": 8.130081300813009e-08,
      "loss": 1.4496,
      "step": 123
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 123,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.370316693556429e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}