trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.0,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.6,
      "learning_rate": 0.0001996129859609174,
      "loss": 2.4955,
      "step": 100
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.00019377521321470805,
      "loss": 1.3769,
      "step": 200
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00018130731609142177,
      "loss": 1.2148,
      "step": 300
    },
    {
      "epoch": 6.4,
      "learning_rate": 0.00016309080374272232,
      "loss": 1.0464,
      "step": 400
    },
    {
      "epoch": 8.0,
      "learning_rate": 0.00014041362570717172,
      "loss": 0.8909,
      "step": 500
    }
  ],
  "max_steps": 1240,
  "num_train_epochs": 20,
  "total_flos": 1.9338439110721536e+16,
  "trial_name": null,
  "trial_params": null
}