Erya4FT / tokenizer_config.json
straybird's picture
Upload 5 files
b9b09f8
raw
history blame contribute delete
559 Bytes
{
"bos_token": "[CLS]",
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": false,
"eos_token": "[EOS]",
"mask_token": "[MASK]",
"model_max_length": 1000000000000000019884624838656,
"name_or_path": "fnlp/cpt-base",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": "vocab/cpt_v3_vocab/special_tokens_map.json",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"tokenizer_file": null,
"unk_token": "[UNK]",
"use_fast": true
}