trainer_state.json

{
  "best_metric": 0.07308927178382874,
  "best_model_checkpoint": "BioM-ELECTRA-Large-SQuAD2-XGHJ/checkpoint-579",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 579,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 0.21314279735088348,
      "eval_runtime": 36.2842,
      "eval_samples_per_second": 42.525,
      "eval_steps_per_second": 5.319,
      "step": 193
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.09692101925611496,
      "eval_runtime": 36.0475,
      "eval_samples_per_second": 42.805,
      "eval_steps_per_second": 5.354,
      "step": 386
    },
    {
      "epoch": 2.5906735751295336,
      "grad_norm": 0.08420258015394211,
      "learning_rate": 2.728842832469776e-06,
      "loss": 0.3808,
      "step": 500
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.07308927178382874,
      "eval_runtime": 36.217,
      "eval_samples_per_second": 42.604,
      "eval_steps_per_second": 5.329,
      "step": 579
    }
  ],
  "logging_steps": 500,
  "max_steps": 579,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 3224238499920384.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}