generation_config.json

{
  "do_sample": true,
  "temperature": 0.7,
  "top_k": 50,
  "top_p": 0.90,
  "typical_p": 0.95,
  "repetition_penalty": 1.2,
  "length_penalty": 1.0,
  "num_beams": 1,
  "num_return_sequences": 1,
  "early_stopping": false,
  "max_new_tokens": 256,
  "min_new_tokens": 16,
  "pad_token_id": 0,
  "bos_token_id": 2,
  "eos_token_id": 3,
  "use_cache": true,
  "return_dict_in_generate": true,
  "output_scores": false,
  "renormalize_logits": false,
  "exponential_decay_length_penalty": null
}