Wiki-Test2 / SAVE_tokenizer_config_SAVE.json
PhysiQuanty's picture
Duplicate from PhysiQuanty/Patent-Test-Radix-65536-AutoTokenizer_FineTune
efda231
raw
history blame contribute delete
750 Bytes
{
"added_tokens_decoder": {
"65536": {
"content": "<BOS>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"65537": {
"content": "<UNK>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
}
},
"bos_token": "<BOS>",
"clean_up_tokenization_spaces": false,
"eos_token": "<EOS>",
"extra_special_tokens": {},
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<EOS>",
"tokenizer_class": "tokenization_binaryllm.BinaryLLMTokenizer",
"unk_token": "<UNK>",
"auto_map": {
"AutoTokenizer": "tokenization_binaryllm.BinaryLLMTokenizer"
}
}