{
  "best_metric": 0.9400261630832187,
  "best_model_checkpoint": "model_saves/xlnet-large_spell_10k_2_p3/checkpoint-268",
  "epoch": 4.0,
  "global_step": 1072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9400261630832187,
      "eval_loss": 0.4545843303203583,
      "eval_runtime": 5.2111,
      "eval_samples_per_second": 840.896,
      "eval_steps_per_second": 6.716,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9395509788794597,
      "eval_loss": 0.463368684053421,
      "eval_runtime": 4.9144,
      "eval_samples_per_second": 891.657,
      "eval_steps_per_second": 7.122,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9397242813537718,
      "eval_loss": 0.5049863457679749,
      "eval_runtime": 4.9181,
      "eval_samples_per_second": 890.999,
      "eval_steps_per_second": 7.117,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.2924,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9397075101465804,
      "eval_loss": 0.5557319521903992,
      "eval_runtime": 4.9099,
      "eval_samples_per_second": 892.485,
      "eval_steps_per_second": 7.128,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.7283799071064064e+16,
      "train_loss": 0.28738758724127245,
      "train_runtime": 454.0424,
      "train_samples_per_second": 1133.286,
      "train_steps_per_second": 8.854
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.7283799071064064e+16,
  "trial_name": null,
  "trial_params": null
}