tokenizer-pl / tokenizer_config.json
suchirsalhan's picture
Final Fix: Correct Metaspace mapping and Unigram scores
b868b13 verified
raw
history blame contribute delete
261 Bytes
{
"backend": "tokenizers",
"bos_token": "<s>",
"clean_up_tokenization_spaces": false,
"eos_token": "</s>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<pad>",
"tokenizer_class": "TokenizersBackend",
"unk_token": "<unk>"
}