run-1/checkpoint-32/trainer_state.json

{
  "best_metric": 0.5523465703971119,
  "best_model_checkpoint": "microsoft/deberta-v3-smallsed_rte-finetuned-rte/run-1/checkpoint-32",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 32,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5523465703971119,
      "eval_loss": 0.6929216384887695,
      "eval_runtime": 1.7369,
      "eval_samples_per_second": 159.476,
      "eval_steps_per_second": 10.363,
      "step": 32
    }
  ],
  "logging_steps": 500,
  "max_steps": 128,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "total_flos": 0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.8381474440323331e-06,
    "num_train_epochs": 4,
    "per_device_train_batch_size": 8,
    "seed": 31
  }
}