trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 1000,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.67,
      "learning_rate": 4.266666666666667e-05,
      "loss": 8.0945,
      "step": 25
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.433333333333333e-05,
      "loss": 7.5595,
      "step": 50
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 7.2051,
      "step": 75
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.7666666666666668e-05,
      "loss": 7.1045,
      "step": 100
    },
    {
      "epoch": 8.33,
      "learning_rate": 9.333333333333334e-06,
      "loss": 7.0282,
      "step": 125
    },
    {
      "epoch": 10.0,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 6.9739,
      "step": 150
    },
    {
      "epoch": 10.0,
      "step": 150,
      "total_flos": 0.0,
      "train_loss": 7.327593892415365,
      "train_runtime": 114.1866,
      "train_samples_per_second": 2.627,
      "train_steps_per_second": 1.314
    }
  ],
  "logging_steps": 25,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}