{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.1928721174004195,
  "global_step": 6000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 86.0421,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.000000000000001e-06,
      "loss": 51.6706,
      "step": 200
    },
    {
      "epoch": 0.21,
      "learning_rate": 6e-06,
      "loss": 41.1349,
      "step": 300
    },
    {
      "epoch": 0.28,
      "learning_rate": 8.000000000000001e-06,
      "loss": 36.5061,
      "step": 400
    },
    {
      "epoch": 0.35,
      "learning_rate": 1e-05,
      "loss": 33.1858,
      "step": 500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2e-05,
      "loss": 30.5206,
      "step": 600
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 28.0073,
      "step": 700
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 26.1939,
      "step": 800
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8e-05,
      "loss": 24.3465,
      "step": 900
    },
    {
      "epoch": 0.7,
      "learning_rate": 2e-05,
      "loss": 22.5759,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 20.9294,
      "step": 1100
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.4e-05,
      "loss": 19.3762,
      "step": 1200
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 17.72,
      "step": 1300
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 15.7901,
      "step": 1400
    },
    {
      "epoch": 1.05,
      "learning_rate": 3e-05,
      "loss": 14.0008,
      "step": 1500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 12.3777,
      "step": 1600
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 10.7261,
      "step": 1700
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.6e-05,
      "loss": 9.1024,
      "step": 1800
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.8e-05,
      "loss": 7.4676,
      "step": 1900
    },
    {
      "epoch": 1.4,
      "learning_rate": 4e-05,
      "loss": 6.6044,
      "step": 2000
    },
    {
      "epoch": 1.4,
      "eval_bleu": 3.8045,
      "eval_em": 0.0,
      "eval_gen_len": 158.8473,
      "eval_loss": 6.171305179595947,
      "eval_runtime": 1556.9786,
      "eval_samples_per_second": 0.774,
      "eval_steps_per_second": 0.097,
      "step": 2000
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.2e-05,
      "loss": 6.0941,
      "step": 2100
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 5.6741,
      "step": 2200
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.600000000000001e-05,
      "loss": 5.4757,
      "step": 2300
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.8e-05,
      "loss": 5.242,
      "step": 2400
    },
    {
      "epoch": 1.75,
      "learning_rate": 5e-05,
      "loss": 5.0108,
      "step": 2500
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.944121591417077e-05,
      "loss": 4.8595,
      "step": 2600
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.888243182834153e-05,
      "loss": 4.695,
      "step": 2700
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.8323647742512295e-05,
      "loss": 4.5706,
      "step": 2800
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.776486365668306e-05,
      "loss": 4.2498,
      "step": 2900
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.720607957085382e-05,
      "loss": 4.1223,
      "step": 3000
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.664729548502459e-05,
      "loss": 4.0181,
      "step": 3100
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.6088511399195353e-05,
      "loss": 3.8722,
      "step": 3200
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.552972731336611e-05,
      "loss": 3.7786,
      "step": 3300
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.497094322753688e-05,
      "loss": 3.6403,
      "step": 3400
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.4412159141707646e-05,
      "loss": 3.5437,
      "step": 3500
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.385337505587841e-05,
      "loss": 3.389,
      "step": 3600
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.329459097004918e-05,
      "loss": 3.2395,
      "step": 3700
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.2735806884219945e-05,
      "loss": 3.1786,
      "step": 3800
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.2177022798390704e-05,
      "loss": 3.0657,
      "step": 3900
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.161823871256147e-05,
      "loss": 3.032,
      "step": 4000
    },
    {
      "epoch": 2.8,
      "eval_bleu": 27.701,
      "eval_em": 0.0,
      "eval_gen_len": 33.9568,
      "eval_loss": 2.904534339904785,
      "eval_runtime": 297.5953,
      "eval_samples_per_second": 4.049,
      "eval_steps_per_second": 0.507,
      "step": 4000
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.105945462673223e-05,
      "loss": 2.8755,
      "step": 4100
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.0500670540903e-05,
      "loss": 2.8396,
      "step": 4200
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.994188645507376e-05,
      "loss": 2.7454,
      "step": 4300
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.938310236924452e-05,
      "loss": 2.5218,
      "step": 4400
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.882431828341529e-05,
      "loss": 2.4895,
      "step": 4500
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.8265534197586055e-05,
      "loss": 2.4554,
      "step": 4600
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.7706750111756815e-05,
      "loss": 2.3573,
      "step": 4700
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.714796602592758e-05,
      "loss": 2.2979,
      "step": 4800
    },
    {
      "epoch": 3.42,
      "learning_rate": 3.658918194009835e-05,
      "loss": 2.1874,
      "step": 4900
    },
    {
      "epoch": 3.49,
      "learning_rate": 3.603039785426911e-05,
      "loss": 2.1803,
      "step": 5000
    },
    {
      "epoch": 3.56,
      "learning_rate": 3.5471613768439874e-05,
      "loss": 2.1553,
      "step": 5100
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.491282968261064e-05,
      "loss": 2.0567,
      "step": 5200
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.4354045596781406e-05,
      "loss": 2.0147,
      "step": 5300
    },
    {
      "epoch": 3.77,
      "learning_rate": 3.379526151095217e-05,
      "loss": 1.9817,
      "step": 5400
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.323647742512294e-05,
      "loss": 1.8843,
      "step": 5500
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.26776933392937e-05,
      "loss": 1.8849,
      "step": 5600
    },
    {
      "epoch": 3.98,
      "learning_rate": 3.2118909253464465e-05,
      "loss": 1.8232,
      "step": 5700
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.156012516763523e-05,
      "loss": 1.6714,
      "step": 5800
    },
    {
      "epoch": 4.12,
      "learning_rate": 3.100134108180599e-05,
      "loss": 1.6047,
      "step": 5900
    },
    {
      "epoch": 4.19,
      "learning_rate": 3.0442556995976757e-05,
      "loss": 1.5893,
      "step": 6000
    },
    {
      "epoch": 4.19,
      "eval_bleu": 57.7509,
      "eval_em": 0.0033,
      "eval_gen_len": 44.132,
      "eval_loss": 1.7326730489730835,
      "eval_runtime": 426.2734,
      "eval_samples_per_second": 2.827,
      "eval_steps_per_second": 0.354,
      "step": 6000
    }
  ],
  "max_steps": 11448,
  "num_train_epochs": 8,
  "total_flos": 3958825611574680.0,
  "trial_name": null,
  "trial_params": null
}