trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.986899563318778,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.75,
      "learning_rate": 8.535533905932739e-06,
      "loss": 4.3872,
      "step": 25
    },
    {
      "epoch": 3.49,
      "learning_rate": 5e-06,
      "loss": 0.3835,
      "step": 50
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.4915,
      "step": 75
    },
    {
      "epoch": 6.99,
      "learning_rate": 0.0,
      "loss": 0.1806,
      "step": 100
    },
    {
      "epoch": 6.99,
      "step": 100,
      "total_flos": 5.186534972986491e+17,
      "train_loss": 1.3607243776321412,
      "train_runtime": 7371.0025,
      "train_samples_per_second": 0.868,
      "train_steps_per_second": 0.014
    }
  ],
  "max_steps": 100,
  "num_train_epochs": 8,
  "total_flos": 5.186534972986491e+17,
  "trial_name": null,
  "trial_params": null
}