run-6/checkpoint-268/trainer_state.json

{
  "best_metric": 0.598778486251831,
  "best_model_checkpoint": "BERT-WMM/run-6/checkpoint-268",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 268,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7352112676056338,
      "eval_loss": 0.6478816270828247,
      "eval_runtime": 2.0455,
      "eval_samples_per_second": 1041.294,
      "eval_steps_per_second": 65.509,
      "step": 134
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7568075117370892,
      "eval_loss": 0.598778486251831,
      "eval_runtime": 2.5526,
      "eval_samples_per_second": 834.457,
      "eval_steps_per_second": 52.496,
      "step": 268
    }
  ],
  "logging_steps": 500,
  "max_steps": 536,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.9530733406603648e-05,
    "num_train_epochs": 4,
    "per_device_train_batch_size": 64,
    "weight_decay": 0.01257257791736148
  }
}