sml-ukr-message-classifier / tokenizer_config.json
jmargutti's picture
Add SetFit model
83f746b
raw
history blame
357 Bytes
{
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"do_lower_case": true,
"eos_token": "</s>",
"mask_token": "<mask>",
"model_max_length": 512,
"pad_token": "<pad>",
"sep_token": "</s>",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "MPNetTokenizer",
"unk_token": "[UNK]"
}