{
  "best_metric": 0.8666666666666667,
  "best_model_checkpoint": "./outputs/xlm-roberta-large-best-model/xnli_16_0.1_0.00001_04-05-23_12-26/checkpoint-73632",
  "epoch": 4.0,
  "global_step": 98176,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 7.978761445522026e-06,
      "loss": 0.548,
      "step": 24544
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8526104417670682,
      "eval_loss": 0.4045264422893524,
      "eval_runtime": 85.4254,
      "eval_samples_per_second": 29.148,
      "eval_steps_per_second": 3.652,
      "step": 24544
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.319174297014683e-06,
      "loss": 0.3739,
      "step": 49088
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8538152610441767,
      "eval_loss": 0.3931969106197357,
      "eval_runtime": 85.4189,
      "eval_samples_per_second": 29.15,
      "eval_steps_per_second": 3.653,
      "step": 49088
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.6595871485073417e-06,
      "loss": 0.2739,
      "step": 73632
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8666666666666667,
      "eval_loss": 0.457296758890152,
      "eval_runtime": 85.1723,
      "eval_samples_per_second": 29.235,
      "eval_steps_per_second": 3.663,
      "step": 73632
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0,
      "loss": 0.2075,
      "step": 98176
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8626506024096385,
      "eval_loss": 0.5882326364517212,
      "eval_runtime": 85.1965,
      "eval_samples_per_second": 29.227,
      "eval_steps_per_second": 3.662,
      "step": 98176
    },
    {
      "epoch": 4.0,
      "step": 98176,
      "total_flos": 1.4638901871267594e+18,
      "train_loss": 0.3508187235567374,
      "train_runtime": 165856.5757,
      "train_samples_per_second": 9.471,
      "train_steps_per_second": 0.592
    }
  ],
  "max_steps": 98176,
  "num_train_epochs": 4,
  "total_flos": 1.4638901871267594e+18,
  "trial_name": null,
  "trial_params": null
}