trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.0,
  "global_step": 1560,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.56,
      "learning_rate": 5e-05,
      "loss": 1.2175,
      "step": 500
    },
    {
      "epoch": 5.13,
      "learning_rate": 5e-05,
      "loss": 0.6497,
      "step": 1000
    },
    {
      "epoch": 7.69,
      "learning_rate": 5e-05,
      "loss": 0.361,
      "step": 1500
    },
    {
      "epoch": 8.0,
      "step": 1560,
      "total_flos": 1.7777139952648192e+16,
      "train_loss": 0.7255945914830917,
      "train_runtime": 8445.2384,
      "train_samples_per_second": 1.473,
      "train_steps_per_second": 0.185
    }
  ],
  "max_steps": 1560,
  "num_train_epochs": 8,
  "total_flos": 1.7777139952648192e+16,
  "trial_name": null,
  "trial_params": null
}