run-0/checkpoint-8/trainer_state.json

{
  "best_metric": 0.5270758122743683,
  "best_model_checkpoint": "microsoft/deberta-v3-smallsed_rte-finetuned-rte/run-0/checkpoint-4",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 8,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5270758122743683,
      "eval_loss": 0.6915086507797241,
      "eval_runtime": 1.9301,
      "eval_samples_per_second": 143.52,
      "eval_steps_per_second": 9.326,
      "step": 4
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5270758122743683,
      "eval_loss": 0.6917890906333923,
      "eval_runtime": 1.768,
      "eval_samples_per_second": 156.678,
      "eval_steps_per_second": 10.181,
      "step": 8
    }
  ],
  "logging_steps": 500,
  "max_steps": 20,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 2.0951184212103965e-05,
    "num_train_epochs": 5,
    "per_device_train_batch_size": 64,
    "seed": 39
  }
}