run-1/checkpoint-527/trainer_state.json

{
  "best_metric": 0.875,
  "best_model_checkpoint": "electra-distilled-sst\\run-1\\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": 66.74252319335938,
      "learning_rate": 0.0002000029100592468,
      "loss": 2.5183,
      "step": 527
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.875,
      "eval_loss": 2.3820443153381348,
      "eval_runtime": 1.4493,
      "eval_samples_per_second": 601.689,
      "eval_steps_per_second": 4.83,
      "step": 527
    }
  ],
  "logging_steps": 500,
  "max_steps": 2635,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 186587234208876.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.15523880635662113,
    "learning_rate": 0.0002496483505476624,
    "num_train_epochs": 5,
    "temperature": 16
  }
}