trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "global_step": 2800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.93,
      "learning_rate": 4.107142857142857e-05,
      "loss": 1.8596,
      "step": 500
    },
    {
      "epoch": 17.86,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 1.3136,
      "step": 1000
    },
    {
      "epoch": 26.79,
      "learning_rate": 2.3214285714285715e-05,
      "loss": 1.0178,
      "step": 1500
    },
    {
      "epoch": 35.71,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.8233,
      "step": 2000
    },
    {
      "epoch": 44.64,
      "learning_rate": 5.357142857142857e-06,
      "loss": 0.7052,
      "step": 2500
    },
    {
      "epoch": 50.0,
      "step": 2800,
      "total_flos": 3.305781348671693e+17,
      "train_loss": 1.0920553370884487,
      "train_runtime": 63007.8759,
      "train_samples_per_second": 17.773,
      "train_steps_per_second": 0.044
    }
  ],
  "max_steps": 2800,
  "num_train_epochs": 50,
  "total_flos": 3.305781348671693e+17,
  "trial_name": null,
  "trial_params": null
}