tokenizer_config.json

{"init_inputs": [], "use_cls_at_first": false, "use_sep_at_last": false, "use_start_token": true, "start_token": "[UNK]", "chinese_converter_style": "t2s"}