trainer_state.json

{
  "best_metric": 0.938813045763034,
  "best_model_checkpoint": "model_saves/electra-large_lemon_5k_4_p3/checkpoint-268",
  "epoch": 4.0,
  "global_step": 1072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.938813045763034,
      "eval_loss": 0.4380951225757599,
      "eval_runtime": 3.3893,
      "eval_samples_per_second": 1292.885,
      "eval_steps_per_second": 10.327,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9382763671329062,
      "eval_loss": 0.45095598697662354,
      "eval_runtime": 10.4113,
      "eval_samples_per_second": 420.891,
      "eval_steps_per_second": 3.362,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9382540055233176,
      "eval_loss": 0.47314006090164185,
      "eval_runtime": 3.3998,
      "eval_samples_per_second": 1288.89,
      "eval_steps_per_second": 10.295,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.3237,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9371918290678563,
      "eval_loss": 0.5063077211380005,
      "eval_runtime": 3.3991,
      "eval_samples_per_second": 1289.183,
      "eval_steps_per_second": 10.297,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.3903119448539136e+16,
      "train_loss": 0.3187360105229847,
      "train_runtime": 321.0232,
      "train_samples_per_second": 1602.875,
      "train_steps_per_second": 12.522
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.3903119448539136e+16,
  "trial_name": null,
  "trial_params": null
}