ijelid-indobertweet / tokenizer_config.json
fathan's picture
Training in progress, epoch 1
c45221e
raw
history blame contribute delete
599 Bytes
{
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": true,
"mask_token": "[MASK]",
"model_max_length": 128,
"name_or_path": "indolem/indobertweet-base-uncased",
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"special_tokens_map_file": "/home/jupyter-145230403/.cache/huggingface/transformers/f4e40cfa6532f9592ed04e256a91ed9b5574d28a8e0ad5e760524e96d034c486.dd8bd9bfd3664b530ea4e645105f557769387b3da9f79bdb55ed556bdd80611d",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]"
}