trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 3003,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 4.167499167499168e-05,
      "loss": 8.825,
      "step": 500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.334998334998335e-05,
      "loss": 7.1815,
      "step": 1000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5024975024975027e-05,
      "loss": 6.9532,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6699966699966702e-05,
      "loss": 6.7979,
      "step": 2000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.374958374958376e-06,
      "loss": 6.7027,
      "step": 2500
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.9950049950049954e-08,
      "loss": 6.7203,
      "step": 3000
    }
  ],
  "max_steps": 3003,
  "num_train_epochs": 3,
  "total_flos": 30316878922464.0,
  "trial_name": null,
  "trial_params": null
}