generation_config.json

{
  "max_new_tokens": 150,
  "temperature": 0.9,
  "top_p": 0.9,
  "do_sample": true,
  "num_beams": 1,
  "early_stopping": true
}