sealion-bert-base / tokenizer_config.json
ngannt's picture
Upload folder using huggingface_hub
426f887
raw
history blame
453 Bytes
{
"add_bos_token": true,
"add_eos_token": true,
"auto_map": {
"AutoTokenizer": ["tokenization_SEA_BPE.SEABPETokenizer", null]
},
"bos_token": "<|bos|>",
"clean_up_tokenization_spaces": false,
"eos_token": "<|eos|>",
"legacy": true,
"mask_token": "<|mask|>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<|pad|>",
"sp_model_kwargs": {},
"tokenizer_class": "SEABPETokenizer",
"unk_token": "<|unk|>"
}