trainer_state.json

{
  "best_metric": 0.6723194718360901,
  "best_model_checkpoint": "test_trainer/checkpoint-2000",
  "epoch": 2.0,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 4.75e-05,
      "loss": 0.7766,
      "step": 500
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.5e-05,
      "loss": 0.7784,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.7116053700447083,
      "eval_runtime": 58.005,
      "eval_samples_per_second": 34.48,
      "eval_steps_per_second": 4.31,
      "step": 1000
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.25e-05,
      "loss": 0.6984,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "learning_rate": 4e-05,
      "loss": 0.7149,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.6723194718360901,
      "eval_runtime": 58.1448,
      "eval_samples_per_second": 34.397,
      "eval_steps_per_second": 4.3,
      "step": 2000
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 10,
  "total_flos": 4209288456812544.0,
  "trial_name": null,
  "trial_params": null
}