trainer_state.json

{
  "best_metric": 2.33189058303833,
  "best_model_checkpoint": "output/dj-artem-artemov/checkpoint-18",
  "epoch": 1.0,
  "global_step": 18,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.28,
      "learning_rate": 0.00011269523002449659,
      "loss": 2.0563,
      "step": 5
    },
    {
      "epoch": 0.56,
      "learning_rate": 5.668773501204858e-05,
      "loss": 1.6138,
      "step": 10
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.190657300387505e-06,
      "loss": 1.6715,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.33189058303833,
      "eval_runtime": 0.5891,
      "eval_samples_per_second": 44.135,
      "eval_steps_per_second": 6.79,
      "step": 18
    }
  ],
  "max_steps": 18,
  "num_train_epochs": 1,
  "total_flos": 18813026304000.0,
  "trial_name": null,
  "trial_params": null
}