English

Not All Attention Is All You Need

Computation and Language 2021-06-02 v3

Abstract

Beyond the success story of pre-trained language models (PrLMs) in recent natural language processing, they are susceptible to over-fitting due to unusual large model size. To this end, dropout serves as a therapy. However, existing methods like random-based, knowledge-based and search-based dropout are more general but less effective onto self-attention based models, which are broadly chosen as the fundamental architecture of PrLMs. In this paper, we propose a novel dropout method named AttendOut to let self-attention empowered PrLMs capable of more robust task-specific tuning. We demonstrate that state-of-the-art models with elaborate training design may achieve much stronger results. We verify the universality of our approach on extensive natural language processing tasks.

Keywords

Cite

@article{arxiv.2104.04692,
  title  = {Not All Attention Is All You Need},
  author = {Hongqiu Wu and Hai Zhao and Min Zhang},
  journal= {arXiv preprint arXiv:2104.04692},
  year   = {2021}
}
R2 v1 2026-06-24T01:01:52.436Z