ShareCaptioner / tokenizer_config.json
chenlin
init
8bd85e0
raw
history blame
398 Bytes
{
"auto_map": {
"AutoTokenizer": [
"tokenization_InternLM_XComposer.InternLMXComposerTokenizer",
null
]
},
"bos_token": "<s>",
"clean_up_tokenization_spaces": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "</s>",
"padding_side": "right",
"tokenizer_class": "InternLMXComposerTokenizer",
"unk_token": "<unk>"
}