Contract-new-tokenizer-roberta-large / tokenizer_config.json
jhn9803's picture
Upload tokenizer
c05a070
{
"bos_token": "[CLS]",
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": false,
"eos_token": "[SEP]",
"mask_token": "[MASK]",
"model_max_length": 512,
"name_or_path": "jhn9803/roberta-large-law-contract-tokenizer",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": "/root/.cache/huggingface/hub/models--klue--roberta-large/snapshots/5193b95701189160c45d02a1033a4ea55bdbe259/special_tokens_map.json",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]"
}