{
  "best_metric": 0.6652947068214417,
  "best_model_checkpoint": "../Modelos/mt0_QG_SQuAD/checkpoint-4000",
  "epoch": 2.923512489154756,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "eval_loss": 1.2921833992004395,
      "eval_runtime": 99.039,
      "eval_samples_per_second": 106.726,
      "eval_steps_per_second": 26.686,
      "step": 50
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.7744602560997009,
      "eval_runtime": 99.0509,
      "eval_samples_per_second": 106.713,
      "eval_steps_per_second": 26.683,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.7465793490409851,
      "eval_runtime": 99.0509,
      "eval_samples_per_second": 106.713,
      "eval_steps_per_second": 26.683,
      "step": 150
    },
    {
      "epoch": 0.15,
      "eval_loss": 0.7375356554985046,
      "eval_runtime": 99.0564,
      "eval_samples_per_second": 106.707,
      "eval_steps_per_second": 26.682,
      "step": 200
    },
    {
      "epoch": 0.18,
      "eval_loss": 0.7277144193649292,
      "eval_runtime": 99.0472,
      "eval_samples_per_second": 106.717,
      "eval_steps_per_second": 26.684,
      "step": 250
    },
    {
      "epoch": 0.22,
      "eval_loss": 0.7252267003059387,
      "eval_runtime": 99.0554,
      "eval_samples_per_second": 106.708,
      "eval_steps_per_second": 26.682,
      "step": 300
    },
    {
      "epoch": 0.26,
      "eval_loss": 0.7177889347076416,
      "eval_runtime": 99.0562,
      "eval_samples_per_second": 106.707,
      "eval_steps_per_second": 26.682,
      "step": 350
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.7163848876953125,
      "eval_runtime": 99.0555,
      "eval_samples_per_second": 106.708,
      "eval_steps_per_second": 26.682,
      "step": 400
    },
    {
      "epoch": 0.33,
      "eval_loss": 0.7115530967712402,
      "eval_runtime": 99.0571,
      "eval_samples_per_second": 106.706,
      "eval_steps_per_second": 26.682,
      "step": 450
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.39083820662768e-05,
      "loss": 1.3919,
      "step": 500
    },
    {
      "epoch": 0.37,
      "eval_loss": 0.7089862823486328,
      "eval_runtime": 99.0105,
      "eval_samples_per_second": 106.756,
      "eval_steps_per_second": 26.694,
      "step": 500
    },
    {
      "epoch": 0.4,
      "eval_loss": 0.7095581889152527,
      "eval_runtime": 99.0746,
      "eval_samples_per_second": 106.687,
      "eval_steps_per_second": 26.677,
      "step": 550
    },
    {
      "epoch": 0.44,
      "eval_loss": 0.7025607228279114,
      "eval_runtime": 99.0556,
      "eval_samples_per_second": 106.708,
      "eval_steps_per_second": 26.682,
      "step": 600
    },
    {
      "epoch": 0.47,
      "eval_loss": 0.7023607492446899,
      "eval_runtime": 99.0561,
      "eval_samples_per_second": 106.707,
      "eval_steps_per_second": 26.682,
      "step": 650
    },
    {
      "epoch": 0.51,
      "eval_loss": 0.6990346312522888,
      "eval_runtime": 99.0585,
      "eval_samples_per_second": 106.705,
      "eval_steps_per_second": 26.681,
      "step": 700
    },
    {
      "epoch": 0.55,
      "eval_loss": 0.698978066444397,
      "eval_runtime": 99.0563,
      "eval_samples_per_second": 106.707,
      "eval_steps_per_second": 26.682,
      "step": 750
    },
    {
      "epoch": 0.58,
      "eval_loss": 0.6978702545166016,
      "eval_runtime": 99.0654,
      "eval_samples_per_second": 106.697,
      "eval_steps_per_second": 26.679,
      "step": 800
    },
    {
      "epoch": 0.62,
      "eval_loss": 0.6929869651794434,
      "eval_runtime": 99.0669,
      "eval_samples_per_second": 106.696,
      "eval_steps_per_second": 26.679,
      "step": 850
    },
    {
      "epoch": 0.66,
      "eval_loss": 0.693868100643158,
      "eval_runtime": 99.063,
      "eval_samples_per_second": 106.7,
      "eval_steps_per_second": 26.68,
      "step": 900
    },
    {
      "epoch": 0.69,
      "eval_loss": 0.6937694549560547,
      "eval_runtime": 99.0627,
      "eval_samples_per_second": 106.7,
      "eval_steps_per_second": 26.68,
      "step": 950
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7816764132553604e-05,
      "loss": 0.7713,
      "step": 1000
    },
    {
      "epoch": 0.73,
      "eval_loss": 0.6868888139724731,
      "eval_runtime": 99.0162,
      "eval_samples_per_second": 106.75,
      "eval_steps_per_second": 26.693,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "eval_loss": 0.68803870677948,
      "eval_runtime": 99.0558,
      "eval_samples_per_second": 106.708,
      "eval_steps_per_second": 26.682,
      "step": 1050
    },
    {
      "epoch": 0.8,
      "eval_loss": 0.6879799365997314,
      "eval_runtime": 99.0081,
      "eval_samples_per_second": 106.759,
      "eval_steps_per_second": 26.695,
      "step": 1100
    },
    {
      "epoch": 0.84,
      "eval_loss": 0.6890503764152527,
      "eval_runtime": 99.0519,
      "eval_samples_per_second": 106.712,
      "eval_steps_per_second": 26.683,
      "step": 1150
    },
    {
      "epoch": 0.88,
      "eval_loss": 0.6867082715034485,
      "eval_runtime": 99.0571,
      "eval_samples_per_second": 106.706,
      "eval_steps_per_second": 26.682,
      "step": 1200
    },
    {
      "epoch": 0.91,
      "eval_loss": 0.685232400894165,
      "eval_runtime": 99.0511,
      "eval_samples_per_second": 106.713,
      "eval_steps_per_second": 26.683,
      "step": 1250
    },
    {
      "epoch": 0.95,
      "eval_loss": 0.6860905885696411,
      "eval_runtime": 99.0563,
      "eval_samples_per_second": 106.707,
      "eval_steps_per_second": 26.682,
      "step": 1300
    },
    {
      "epoch": 0.99,
      "eval_loss": 0.6822260022163391,
      "eval_runtime": 99.0589,
      "eval_samples_per_second": 106.704,
      "eval_steps_per_second": 26.681,
      "step": 1350
    },
    {
      "epoch": 1.02,
      "eval_loss": 0.6809637546539307,
      "eval_runtime": 99.0612,
      "eval_samples_per_second": 106.702,
      "eval_steps_per_second": 26.68,
      "step": 1400
    },
    {
      "epoch": 1.06,
      "eval_loss": 0.6814208030700684,
      "eval_runtime": 99.0551,
      "eval_samples_per_second": 106.708,
      "eval_steps_per_second": 26.682,
      "step": 1450
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.172514619883041e-05,
      "loss": 0.7477,
      "step": 1500
    },
    {
      "epoch": 1.1,
      "eval_loss": 0.6810862421989441,
      "eval_runtime": 99.0227,
      "eval_samples_per_second": 106.743,
      "eval_steps_per_second": 26.691,
      "step": 1500
    },
    {
      "epoch": 1.13,
      "eval_loss": 0.67988121509552,
      "eval_runtime": 99.0581,
      "eval_samples_per_second": 106.705,
      "eval_steps_per_second": 26.681,
      "step": 1550
    },
    {
      "epoch": 1.17,
      "eval_loss": 0.6793538928031921,
      "eval_runtime": 99.0598,
      "eval_samples_per_second": 106.703,
      "eval_steps_per_second": 26.681,
      "step": 1600
    },
    {
      "epoch": 1.21,
      "eval_loss": 0.6803789734840393,
      "eval_runtime": 99.0507,
      "eval_samples_per_second": 106.713,
      "eval_steps_per_second": 26.683,
      "step": 1650
    },
    {
      "epoch": 1.24,
      "eval_loss": 0.6781566739082336,
      "eval_runtime": 99.0472,
      "eval_samples_per_second": 106.717,
      "eval_steps_per_second": 26.684,
      "step": 1700
    },
    {
      "epoch": 1.28,
      "eval_loss": 0.6781692504882812,
      "eval_runtime": 99.0434,
      "eval_samples_per_second": 106.721,
      "eval_steps_per_second": 26.685,
      "step": 1750
    },
    {
      "epoch": 1.32,
      "eval_loss": 0.6762599945068359,
      "eval_runtime": 99.0531,
      "eval_samples_per_second": 106.71,
      "eval_steps_per_second": 26.683,
      "step": 1800
    },
    {
      "epoch": 1.35,
      "eval_loss": 0.6762254238128662,
      "eval_runtime": 99.0448,
      "eval_samples_per_second": 106.719,
      "eval_steps_per_second": 26.685,
      "step": 1850
    },
    {
      "epoch": 1.39,
      "eval_loss": 0.6738188862800598,
      "eval_runtime": 99.0427,
      "eval_samples_per_second": 106.722,
      "eval_steps_per_second": 26.685,
      "step": 1900
    },
    {
      "epoch": 1.43,
      "eval_loss": 0.6767028570175171,
      "eval_runtime": 99.0459,
      "eval_samples_per_second": 106.718,
      "eval_steps_per_second": 26.685,
      "step": 1950
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5633528265107216e-05,
      "loss": 0.7251,
      "step": 2000
    },
    {
      "epoch": 1.46,
      "eval_loss": 0.6748691201210022,
      "eval_runtime": 98.9904,
      "eval_samples_per_second": 106.778,
      "eval_steps_per_second": 26.7,
      "step": 2000
    },
    {
      "epoch": 1.5,
      "eval_loss": 0.6725715398788452,
      "eval_runtime": 99.0539,
      "eval_samples_per_second": 106.71,
      "eval_steps_per_second": 26.682,
      "step": 2050
    },
    {
      "epoch": 1.53,
      "eval_loss": 0.676404595375061,
      "eval_runtime": 99.0537,
      "eval_samples_per_second": 106.71,
      "eval_steps_per_second": 26.683,
      "step": 2100
    },
    {
      "epoch": 1.57,
      "eval_loss": 0.6735058426856995,
      "eval_runtime": 99.0549,
      "eval_samples_per_second": 106.709,
      "eval_steps_per_second": 26.682,
      "step": 2150
    },
    {
      "epoch": 1.61,
      "eval_loss": 0.6737805604934692,
      "eval_runtime": 99.0492,
      "eval_samples_per_second": 106.715,
      "eval_steps_per_second": 26.684,
      "step": 2200
    },
    {
      "epoch": 1.64,
      "eval_loss": 0.6724738478660583,
      "eval_runtime": 99.0578,
      "eval_samples_per_second": 106.705,
      "eval_steps_per_second": 26.681,
      "step": 2250
    },
    {
      "epoch": 1.68,
      "eval_loss": 0.672220766544342,
      "eval_runtime": 99.0481,
      "eval_samples_per_second": 106.716,
      "eval_steps_per_second": 26.684,
      "step": 2300
    },
    {
      "epoch": 1.72,
      "eval_loss": 0.6724982857704163,
      "eval_runtime": 99.0532,
      "eval_samples_per_second": 106.71,
      "eval_steps_per_second": 26.683,
      "step": 2350
    },
    {
      "epoch": 1.75,
      "eval_loss": 0.6707101464271545,
      "eval_runtime": 99.0487,
      "eval_samples_per_second": 106.715,
      "eval_steps_per_second": 26.684,
      "step": 2400
    },
    {
      "epoch": 1.79,
      "eval_loss": 0.6724550127983093,
      "eval_runtime": 99.0502,
      "eval_samples_per_second": 106.714,
      "eval_steps_per_second": 26.683,
      "step": 2450
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9541910331384016e-05,
      "loss": 0.7154,
      "step": 2500
    },
    {
      "epoch": 1.83,
      "eval_loss": 0.6708072423934937,
      "eval_runtime": 98.996,
      "eval_samples_per_second": 106.772,
      "eval_steps_per_second": 26.698,
      "step": 2500
    },
    {
      "epoch": 1.86,
      "eval_loss": 0.6708223223686218,
      "eval_runtime": 99.0485,
      "eval_samples_per_second": 106.715,
      "eval_steps_per_second": 26.684,
      "step": 2550
    },
    {
      "epoch": 1.9,
      "eval_loss": 0.6708201169967651,
      "eval_runtime": 99.0348,
      "eval_samples_per_second": 106.73,
      "eval_steps_per_second": 26.688,
      "step": 2600
    },
    {
      "epoch": 1.94,
      "eval_loss": 0.6681162714958191,
      "eval_runtime": 99.0391,
      "eval_samples_per_second": 106.726,
      "eval_steps_per_second": 26.686,
      "step": 2650
    },
    {
      "epoch": 1.97,
      "eval_loss": 0.6703566908836365,
      "eval_runtime": 99.0413,
      "eval_samples_per_second": 106.723,
      "eval_steps_per_second": 26.686,
      "step": 2700
    },
    {
      "epoch": 2.01,
      "eval_loss": 0.6685038805007935,
      "eval_runtime": 99.0472,
      "eval_samples_per_second": 106.717,
      "eval_steps_per_second": 26.684,
      "step": 2750
    },
    {
      "epoch": 2.05,
      "eval_loss": 0.6697913408279419,
      "eval_runtime": 99.0345,
      "eval_samples_per_second": 106.731,
      "eval_steps_per_second": 26.688,
      "step": 2800
    },
    {
      "epoch": 2.08,
      "eval_loss": 0.6686891913414001,
      "eval_runtime": 99.036,
      "eval_samples_per_second": 106.729,
      "eval_steps_per_second": 26.687,
      "step": 2850
    },
    {
      "epoch": 2.12,
      "eval_loss": 0.6696143746376038,
      "eval_runtime": 99.0322,
      "eval_samples_per_second": 106.733,
      "eval_steps_per_second": 26.688,
      "step": 2900
    },
    {
      "epoch": 2.16,
      "eval_loss": 0.6674590110778809,
      "eval_runtime": 99.0384,
      "eval_samples_per_second": 106.726,
      "eval_steps_per_second": 26.687,
      "step": 2950
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3450292397660819e-05,
      "loss": 0.7119,
      "step": 3000
    },
    {
      "epoch": 2.19,
      "eval_loss": 0.6682441830635071,
      "eval_runtime": 98.9894,
      "eval_samples_per_second": 106.779,
      "eval_steps_per_second": 26.7,
      "step": 3000
    },
    {
      "epoch": 2.23,
      "eval_loss": 0.6684902906417847,
      "eval_runtime": 99.045,
      "eval_samples_per_second": 106.719,
      "eval_steps_per_second": 26.685,
      "step": 3050
    },
    {
      "epoch": 2.27,
      "eval_loss": 0.6682748198509216,
      "eval_runtime": 99.0366,
      "eval_samples_per_second": 106.728,
      "eval_steps_per_second": 26.687,
      "step": 3100
    },
    {
      "epoch": 2.3,
      "eval_loss": 0.6677591800689697,
      "eval_runtime": 99.0364,
      "eval_samples_per_second": 106.728,
      "eval_steps_per_second": 26.687,
      "step": 3150
    },
    {
      "epoch": 2.34,
      "eval_loss": 0.666348397731781,
      "eval_runtime": 99.0376,
      "eval_samples_per_second": 106.727,
      "eval_steps_per_second": 26.687,
      "step": 3200
    },
    {
      "epoch": 2.38,
      "eval_loss": 0.6670761704444885,
      "eval_runtime": 99.0329,
      "eval_samples_per_second": 106.732,
      "eval_steps_per_second": 26.688,
      "step": 3250
    },
    {
      "epoch": 2.41,
      "eval_loss": 0.6670991778373718,
      "eval_runtime": 99.0293,
      "eval_samples_per_second": 106.736,
      "eval_steps_per_second": 26.689,
      "step": 3300
    },
    {
      "epoch": 2.45,
      "eval_loss": 0.6664704084396362,
      "eval_runtime": 99.0352,
      "eval_samples_per_second": 106.73,
      "eval_steps_per_second": 26.687,
      "step": 3350
    },
    {
      "epoch": 2.49,
      "eval_loss": 0.666100263595581,
      "eval_runtime": 99.0398,
      "eval_samples_per_second": 106.725,
      "eval_steps_per_second": 26.686,
      "step": 3400
    },
    {
      "epoch": 2.52,
      "eval_loss": 0.6673260927200317,
      "eval_runtime": 99.0455,
      "eval_samples_per_second": 106.719,
      "eval_steps_per_second": 26.685,
      "step": 3450
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.358674463937622e-06,
      "loss": 0.7062,
      "step": 3500
    },
    {
      "epoch": 2.56,
      "eval_loss": 0.666333794593811,
      "eval_runtime": 98.9963,
      "eval_samples_per_second": 106.772,
      "eval_steps_per_second": 26.698,
      "step": 3500
    },
    {
      "epoch": 2.59,
      "eval_loss": 0.666863739490509,
      "eval_runtime": 99.0318,
      "eval_samples_per_second": 106.733,
      "eval_steps_per_second": 26.688,
      "step": 3550
    },
    {
      "epoch": 2.63,
      "eval_loss": 0.6672132611274719,
      "eval_runtime": 99.0274,
      "eval_samples_per_second": 106.738,
      "eval_steps_per_second": 26.69,
      "step": 3600
    },
    {
      "epoch": 2.67,
      "eval_loss": 0.6660241484642029,
      "eval_runtime": 99.0376,
      "eval_samples_per_second": 106.727,
      "eval_steps_per_second": 26.687,
      "step": 3650
    },
    {
      "epoch": 2.7,
      "eval_loss": 0.6664428114891052,
      "eval_runtime": 99.0361,
      "eval_samples_per_second": 106.729,
      "eval_steps_per_second": 26.687,
      "step": 3700
    },
    {
      "epoch": 2.74,
      "eval_loss": 0.6664352416992188,
      "eval_runtime": 99.0223,
      "eval_samples_per_second": 106.744,
      "eval_steps_per_second": 26.691,
      "step": 3750
    },
    {
      "epoch": 2.78,
      "eval_loss": 0.6668286323547363,
      "eval_runtime": 99.0307,
      "eval_samples_per_second": 106.735,
      "eval_steps_per_second": 26.689,
      "step": 3800
    },
    {
      "epoch": 2.81,
      "eval_loss": 0.6657348275184631,
      "eval_runtime": 99.0283,
      "eval_samples_per_second": 106.737,
      "eval_steps_per_second": 26.689,
      "step": 3850
    },
    {
      "epoch": 2.85,
      "eval_loss": 0.6659272313117981,
      "eval_runtime": 99.0306,
      "eval_samples_per_second": 106.735,
      "eval_steps_per_second": 26.689,
      "step": 3900
    },
    {
      "epoch": 2.89,
      "eval_loss": 0.6652637720108032,
      "eval_runtime": 99.0292,
      "eval_samples_per_second": 106.736,
      "eval_steps_per_second": 26.689,
      "step": 3950
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2670565302144249e-06,
      "loss": 0.6933,
      "step": 4000
    },
    {
      "epoch": 2.92,
      "eval_loss": 0.6652947068214417,
      "eval_runtime": 98.9797,
      "eval_samples_per_second": 106.79,
      "eval_steps_per_second": 26.702,
      "step": 4000
    }
  ],
  "max_steps": 4104,
  "num_train_epochs": 3,
  "total_flos": 3.078108528813015e+17,
  "trial_name": null,
  "trial_params": null
}