trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 11.0,
  "eval_steps": 500,
  "global_step": 2750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.0,
      "grad_norm": 8.074329376220703,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.341,
      "step": 500
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7640824913978577,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 0.0878,
      "step": 1000
    },
    {
      "epoch": 6.0,
      "grad_norm": 69.4479751586914,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.0308,
      "step": 1500
    },
    {
      "epoch": 8.0,
      "grad_norm": 61.59275817871094,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.0136,
      "step": 2000
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.014042490161955357,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.0051,
      "step": 2500
    },
    {
      "epoch": 11.0,
      "step": 2750,
      "total_flos": 8.198945748287078e+16,
      "train_loss": 0.0872608635859056,
      "train_runtime": 5767.3328,
      "train_samples_per_second": 15.255,
      "train_steps_per_second": 0.477
    }
  ],
  "logging_steps": 500,
  "max_steps": 2750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 11,
  "save_steps": 500,
  "total_flos": 8.198945748287078e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}