Mask More and Mask Later: Efficient Pre-training of Masked Language Models by Disentangling the [MASK] Token

Liao, Baohao; Thulke, David; Hewavitharana, Sanjika; Ney, Hermann; Monz, Christof

doi:10.18653/v1/2022.findings-emnlp.106

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A similar approach has been explored by masked autoencoders in vision , where 75% of the input patches are masked and removed from the input of the heavy encoder to achieve a 4.1× speedup. Recently, Liao et al (2022) have applied these architectural improvements to natural language pre-training, and together with a high masking rate can accelerate MLM by a third of the pre-training budget.…”

Section: Conclusion and Discussionmentioning

confidence: 99%

Should You Mask 15% in Masked Language Modeling?

Wettig,

Gao,

Zhong

et al. 2023

Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics

View full text Add to dashboard Cite

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 harder; 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. 1 * The first two authors contributed equally. 1 Our code and pre-trained models are publicly available at https://github.com/princeton-nlp/DinkyTrain.

show abstract

Section: Conclusion and Discussionmentioning

confidence: 99%