HELP NEEDED Mask Token in SentencePiece tokenizer HELP NEEDED

[debrupf2946]

T5 using sentencepiece-tokenizer ,I want to perform masked language modeling,
In SentencePiece only There are only BOS,EOS,UNK,PAD so no mask token ,so how to create and add a mask token when the model is called from preteset

[taku910]

See https://github.com/google/sentencepiece/blob/master/doc/special_symbols.md
https://github.com/google/sentencepiece/blob/master/python/sentencepiece_python_module_example.ipynb