timex-thyme-colon / tokenizer_config.json
Tim Miller
Update for cnlpt v0.6.0.
9a45759
raw
history blame contribute delete
571 Bytes
{
"add_prefix_space": true,
"additional_special_tokens": [
"<e>",
"</e>",
"<a1>",
"</a1>",
"<a2>",
"</a2>",
"<cr>",
"<neg>"
],
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": true,
"mask_token": "[MASK]",
"name_or_path": "microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": null,
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]"
}