pretrained_models/chinese-roberta-wwm-ext/tokenizer_config.json

{"init_inputs": []}