trainer_state.json

{
  "best_metric": 0.99,
  "best_model_checkpoint": "imagenet-tiger/checkpoint-50",
  "epoch": 2.0,
  "eval_steps": 20,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "grad_norm": 61.073822021484375,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.8141,
      "step": 20
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.95,
      "eval_loss": 0.2319546490907669,
      "eval_runtime": 73.3596,
      "eval_samples_per_second": 1.363,
      "eval_steps_per_second": 0.341,
      "step": 25
    },
    {
      "epoch": 1.6,
      "grad_norm": 31.59669303894043,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.7293,
      "step": 40
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.09698653966188431,
      "eval_runtime": 71.2695,
      "eval_samples_per_second": 1.403,
      "eval_steps_per_second": 0.351,
      "step": 50
    },
    {
      "epoch": 2.0,
      "step": 50,
      "total_flos": 1.56304031514624e+16,
      "train_loss": 0.6391802191734314,
      "train_runtime": 660.382,
      "train_samples_per_second": 0.303,
      "train_steps_per_second": 0.076
    }
  ],
  "logging_steps": 20,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 20,
  "total_flos": 1.56304031514624e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}