run-4/checkpoint-527/trainer_state.json

{
  "best_metric": 0.7717889908256881,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-4/checkpoint-527",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 527,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 6.212067127227783,
      "learning_rate": 1.7783082729779372e-05,
      "loss": 1.8496,
      "step": 527
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7717889908256881,
      "eval_loss": 1.3957021236419678,
      "eval_runtime": 2.8127,
      "eval_samples_per_second": 310.024,
      "eval_steps_per_second": 2.489,
      "step": 527
    }
  ],
  "logging_steps": 500,
  "max_steps": 3162,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 8057768762220.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.6704103438400755,
    "learning_rate": 2.1339699275735247e-05,
    "num_train_epochs": 6,
    "temperature": 15
  }
}