spam-small-data-bert-base-10e / tokenizer_config.json
jeongah's picture
Upload tokenizer
5f98c37
raw
history blame
525 Bytes
{
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": false,
"mask_token": "[MASK]",
"model_max_length": 512,
"name_or_path": "klue/bert-base",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": "/opt/ml/.cache/huggingface/hub/models--klue--bert-base/snapshots/34b965303f98bc5214daca7f76b7fb82d2dc6183/special_tokens_map.json",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]"
}