checkpoint-3000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.644468313641246,
  "global_step": 3000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 6.10123648e-05,
      "loss": 1.2324,
      "step": 500
    },
    {
      "epoch": 0.21,
      "learning_rate": 5.472524835555556e-05,
      "loss": 0.7401,
      "step": 1000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.789142613333333e-05,
      "loss": 0.6104,
      "step": 1500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.105760391111111e-05,
      "loss": 0.5478,
      "step": 2000
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.4223781688888885e-05,
      "loss": 0.4784,
      "step": 2500
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.7389959466666668e-05,
      "loss": 0.4348,
      "step": 3000
    }
  ],
  "max_steps": 5000,
  "num_train_epochs": 2,
  "total_flos": 1.385209921536e+19,
  "trial_name": null,
  "trial_params": null
}