trainer_state.json

{
  "best_metric": 0.937442502299908,
  "best_model_checkpoint": "data/train-test/roberta-large-output//model/checkpoint-232",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 232,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": null,
      "eval_f1": 0.902638762511374,
      "eval_loss": 0.05000825226306915,
      "eval_precision": 0.8928892889288929,
      "eval_recall": 0.9126034958601656,
      "eval_runtime": 4.4114,
      "eval_samples_per_second": 219.883,
      "eval_steps_per_second": 7.027,
      "step": 116
    },
    {
      "epoch": 2.0,
      "eval_accuracy": null,
      "eval_f1": 0.9276285844333181,
      "eval_loss": 0.03446832671761513,
      "eval_precision": 0.918018018018018,
      "eval_recall": 0.937442502299908,
      "eval_runtime": 4.4048,
      "eval_samples_per_second": 220.215,
      "eval_steps_per_second": 7.038,
      "step": 232
    }
  ],
  "logging_steps": 500,
  "max_steps": 232,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 853475550077862.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}