Should You Mask 15% in Masked Language Modeling?

Alexander Wettig, Tianyu Gao, Zexuan Zhong, Danqi Chen


Abstract
Masked language models (MLMs) conventionally mask 15% of tokens due to the belief that more masking would leave insufficient context to learn good representations; this masking rate has been widely used, regardless of model sizes or masking strategies. In this work, we revisit this important choice of MLM pre-training. We first establish that 15% is not universally optimal, and larger models should adopt a higher masking rate. Specifically, we find that masking 40% outperforms 15% for BERT-large size models on GLUE and SQuAD. Interestingly, an extremely high masking rate of 80% can still preserve 95% fine-tuning performance and most of the accuracy in linguistic probing, challenging the conventional wisdom about the role of the masking rate. We then examine the interplay between masking rates and masking strategies and find that uniform masking requires a higher masking rate compared to sophisticated masking strategies such as span or PMI masking. Finally, we argue that increasing the masking rate has two distinct effects: it leads to more corruption, which makes the prediction task more difficult; it also enables more predictions, which benefits optimization. Using this framework, we revisit BERT’s 80-10-10 corruption strategy. Together, our results contribute to a better understanding of MLM pre-training.
Anthology ID:
2023.eacl-main.217
Volume:
Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics
Month:
May
Year:
2023
Address:
Dubrovnik, Croatia
Editors:
Andreas Vlachos, Isabelle Augenstein
Venue:
EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2985–3000
Language:
URL:
https://aclanthology.org/2023.eacl-main.217
DOI:
10.18653/v1/2023.eacl-main.217
Bibkey:
Cite (ACL):
Alexander Wettig, Tianyu Gao, Zexuan Zhong, and Danqi Chen. 2023. Should You Mask 15% in Masked Language Modeling?. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2985–3000, Dubrovnik, Croatia. Association for Computational Linguistics.
Cite (Informal):
Should You Mask 15% in Masked Language Modeling? (Wettig et al., EACL 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.eacl-main.217.pdf
Video:
 https://aclanthology.org/2023.eacl-main.217.mp4