quantization_model/generation_config.json

{
  "_from_model_config": true,
  "bos_token_id": 50256,
  "eos_token_id": 50256,
  "transformers_version": "4.49.0"
}