{
  "model_max_length": 8192,
  "tokenizer_class": "QWenVLTokenizer"
}