trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.958904109589041,
  "eval_steps": 500,
  "global_step": 54,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.55,
      "grad_norm": 1.638972731060276,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 1.0908,
      "step": 10
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0710882452469266,
      "learning_rate": 1.608761429008721e-05,
      "loss": 0.9545,
      "step": 20
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9859220972082218,
      "learning_rate": 1e-05,
      "loss": 0.8724,
      "step": 30
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1753998705017632,
      "learning_rate": 3.912385709912794e-06,
      "loss": 0.8427,
      "step": 40
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.1336143023825,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 0.8133,
      "step": 50
    },
    {
      "epoch": 2.96,
      "step": 54,
      "total_flos": 9.004999286592635e+17,
      "train_loss": 0.9069184638835766,
      "train_runtime": 2207.3234,
      "train_samples_per_second": 12.543,
      "train_steps_per_second": 0.024
    }
  ],
  "logging_steps": 10,
  "max_steps": 54,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "total_flos": 9.004999286592635e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}