trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 29.7196261682243,
  "eval_steps": 500,
  "global_step": 1590,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.35,
      "learning_rate": 3.4276729559748424e-05,
      "loss": 1.6882,
      "step": 500
    },
    {
      "epoch": 18.69,
      "learning_rate": 1.8553459119496856e-05,
      "loss": 0.9858,
      "step": 1000
    },
    {
      "epoch": 28.04,
      "learning_rate": 2.830188679245283e-06,
      "loss": 0.8402,
      "step": 1500
    },
    {
      "epoch": 29.72,
      "step": 1590,
      "total_flos": 4.438724285646766e+17,
      "train_loss": 1.150932288019912,
      "train_runtime": 3015.3294,
      "train_samples_per_second": 8.457,
      "train_steps_per_second": 0.527
    }
  ],
  "logging_steps": 500,
  "max_steps": 1590,
  "num_train_epochs": 30,
  "save_steps": 200,
  "total_flos": 4.438724285646766e+17,
  "trial_name": null,
  "trial_params": null
}