wangchanglm-7.5B-sft-en-8bit / tokenizer_config.json
wannaphong's picture
Upload tokenizer
caaa15d
{
"additional_special_tokens": [
"<madeupword0>",
"<madeupword1>",
"<madeupword2>",
"<madeupword3>",
"<madeupword4>",
"<madeupword5>",
"<madeupword6>"
],
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<pad>",
"sep_token": "</s>",
"sp_model_kwargs": {},
"tokenizer_class": "XGLMTokenizer",
"unk_token": "<unk>"
}