trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.99797434166104,
  "global_step": 2220,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.68,
      "learning_rate": 3.873873873873874e-05,
      "loss": 6.1827,
      "step": 500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7477477477477483e-05,
      "loss": 5.3106,
      "step": 1000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6216216216216218e-05,
      "loss": 5.1566,
      "step": 1500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.954954954954955e-06,
      "loss": 5.057,
      "step": 2000
    },
    {
      "epoch": 3.0,
      "step": 2220,
      "total_flos": 4640546488320000.0,
      "train_loss": 5.383951053103885,
      "train_runtime": 1331.1447,
      "train_samples_per_second": 6.675,
      "train_steps_per_second": 1.668
    }
  ],
  "max_steps": 2220,
  "num_train_epochs": 3,
  "total_flos": 4640546488320000.0,
  "trial_name": null,
  "trial_params": null
}