Стандартная маскировка на уровне символов
method
ID: standardcharacterlevelmasking
Related Papers
“Is Whole Word Masking Always Better for Chinese BERT?”: Probing on Chinese Grammatical Error Correction
AbstractWhole word masking (WWM), which masks all subwords corresponding to a word at once, makes a better English BERT model. For the Chinese language, however, there is no subword because …...