DeepSilence-tokenizer / tokenizer_config.json
iDSLR's picture
μ‚¬μš©μž μ •μ˜ 특수 토큰 제거 : ν•΄λ‹Ή 토큰을 ν•™μŠ΅ν•˜λŠ” 것은 맀우 λΉ„νš¨μœ¨μ .
10c6436 verified
{
"add_bos_token": false,
"add_prefix_space": false,
"added_tokens_decoder": {
"50256": {
"content": "$~^eos$",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": true
},
"50257": {
"content": "$~^bos$",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"50258": {
"content": "$~^pad$",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"50259": {
"content": "$~^unk$",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
}
},
"bos_token": "$~^bos$",
"clean_up_tokenization_spaces": true,
"eos_token": "$~^eos$",
"errors": "replace",
"model_max_length": 8192,
"pad_token": "$~^pad$",
"tokenizer_class": "GPT2Tokenizer",
"unk_token": "$~^unk$"
}