generation_config.json

{
  "max_new_tokens": 256,
  "do_sample": true,
  "temperature": 0.8,
  "top_p": 0.95,
  "top_k": 50,
  "repetition_penalty": 1.1,
  "eos_token_id": 50256,
  "pad_token_id": 50256
}