tokenizer_config.json

{"do_lower_case": false, "max_len": 512, "additional_special_tokens": ["<REL_SEP>", "[[", "]]", "<<", ">>"]}