trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 2070,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.72,
      "learning_rate": 3.7922705314009665e-05,
      "loss": 2.2113,
      "step": 500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5845410628019323e-05,
      "loss": 1.4796,
      "step": 1000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3768115942028985e-05,
      "loss": 1.226,
      "step": 1500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6908212560386474e-06,
      "loss": 1.0334,
      "step": 2000
    },
    {
      "epoch": 3.0,
      "step": 2070,
      "total_flos": 590265116000256.0,
      "train_loss": 1.4712043319923291,
      "train_runtime": 392.2574,
      "train_samples_per_second": 21.093,
      "train_steps_per_second": 5.277
    }
  ],
  "max_steps": 2070,
  "num_train_epochs": 3,
  "total_flos": 590265116000256.0,
  "trial_name": null,
  "trial_params": null
}