trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.5936001019904067,
  "eval_steps": 50000,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "learning_rate": 4.9920355054102726e-05,
      "loss": 1.1412,
      "step": 50000
    },
    {
      "epoch": 0.8,
      "eval_loss": 1.011365294456482,
      "eval_runtime": 2006.5534,
      "eval_samples_per_second": 111.193,
      "eval_steps_per_second": 1.738,
      "step": 50000
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.992035346050262e-05,
      "loss": 1.0123,
      "step": 100000
    },
    {
      "epoch": 1.59,
      "eval_loss": 0.9693423509597778,
      "eval_runtime": 1995.4615,
      "eval_samples_per_second": 111.811,
      "eval_steps_per_second": 1.747,
      "step": 100000
    }
  ],
  "logging_steps": 50000,
  "max_steps": 31375500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 500,
  "save_steps": 50000,
  "total_flos": 2.0903329898496e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}