{
    "add_prefix_space": false,
    "model_max_length": 256,
    "tokenizer_class": "PreTrainedTokenizerFast"
}