sarashina1-7b / tokenizer_config.json
kajyuuen's picture
Use FastTokenizer
725aab8
raw
history blame
400 Bytes
{
"extra_ids": 0,
"do_lower_case": false,
"keep_accents": true,
"bos_token": "<s>",
"eos_token": "</s>",
"unk_token": "<unk>",
"pad_token": "<pad>",
"mask_token": "<mask>",
"cls_token": "<cls>",
"sep_token": "<sep>",
"padding_side": "left",
"sp_model_kwargs": {},
"special_tokens_map_file": null,
"tokenizer_class": "PreTrainedTokenizerFast"
}