trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.9447731755424065,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.49,
      "learning_rate": 4.3836291913214994e-05,
      "loss": 4.0421,
      "step": 500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.767258382642998e-05,
      "loss": 2.5829,
      "step": 1000
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.150887573964497e-05,
      "loss": 2.26,
      "step": 1500
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.5345167652859964e-05,
      "loss": 2.0942,
      "step": 2000
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.918145956607495e-05,
      "loss": 2.0008,
      "step": 2500
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.3017751479289941e-05,
      "loss": 1.9267,
      "step": 3000
    },
    {
      "epoch": 3.45,
      "learning_rate": 6.854043392504931e-06,
      "loss": 1.8785,
      "step": 3500
    },
    {
      "epoch": 3.94,
      "learning_rate": 6.903353057199211e-07,
      "loss": 1.8417,
      "step": 4000
    }
  ],
  "max_steps": 4056,
  "num_train_epochs": 4,
  "total_flos": 7368941720003490.0,
  "trial_name": null,
  "trial_params": null
}