trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.704225352112676,
  "eval_steps": 500,
  "global_step": 24,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.56,
      "grad_norm": 0.4201306104660034,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 1.8476,
      "step": 5
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.49071383476257324,
      "learning_rate": 3.147047612756302e-05,
      "loss": 1.821,
      "step": 10
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.5262499451637268,
      "learning_rate": 1.5432914190872757e-05,
      "loss": 1.7599,
      "step": 15
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.6670337319374084,
      "learning_rate": 3.3493649053890326e-06,
      "loss": 1.737,
      "step": 20
    },
    {
      "epoch": 2.7,
      "step": 24,
      "total_flos": 5494718507384832.0,
      "train_loss": 1.778053641319275,
      "train_runtime": 108.8546,
      "train_samples_per_second": 3.886,
      "train_steps_per_second": 0.22
    }
  ],
  "logging_steps": 5,
  "max_steps": 24,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 5494718507384832.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}