trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.683363148479428,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.89,
      "learning_rate": 1.403697078115683e-05,
      "loss": 2.5258,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.843708872795105,
      "eval_runtime": 56.6033,
      "eval_samples_per_second": 39.485,
      "eval_steps_per_second": 2.473,
      "step": 559
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.073941562313657e-06,
      "loss": 1.7826,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.7421672344207764,
      "eval_runtime": 56.7224,
      "eval_samples_per_second": 39.402,
      "eval_steps_per_second": 2.468,
      "step": 1118
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.110912343470483e-06,
      "loss": 1.6127,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 1677,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4700206060167168.0,
  "trial_name": null,
  "trial_params": null
}