quantized/gpt2/default/generation_config.json

{
  "bos_token_id": 50256,
  "eos_token_id": 50256,
  "transformers_version": "4.26.0.dev0",
  "_from_model_config": true
}