ملف الباحث
Nafis Sadeq
ورقة واحدة في مجموعة PaperMetrix
المنشورات
أوراق هذا المؤلف
-
InforMask: Unsupervised Informative Masking for Language Model Pretraining
2022 · arXiv (Cornell University)
Masked language modeling is widely used for pretraining large language models for natural language understanding (NLU). However, random masking is suboptimal, allocating an equal masking rate for all tokens. In this paper, we propose InforMask, …