masked-lm-tpu / tokenizer_config.json
luiscunhacsc's picture
Training in progress epoch 0
3d1263b
raw
history blame contribute delete
523 Bytes
{
"bos_token": "[CLS]",
"clean_up_tokenization_spaces": true,
"cls_token": "[CLS]",
"do_lower_case": true,
"eos_token": "[SEP]",
"keep_accents": false,
"mask_token": {
"__type": "AddedToken",
"content": "[MASK]",
"lstrip": true,
"normalized": false,
"rstrip": false,
"single_word": false
},
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<pad>",
"remove_space": true,
"sep_token": "[SEP]",
"tokenizer_class": "AlbertTokenizer",
"unk_token": "<unk>"
}