跳到主要导航 跳到搜索 跳到主要内容

Improving Sequence-to-Sequence Pre-training via Sequence Span Rewriting

  • Wangchunshu Zhou*
  • , Tao Ge
  • , Canwen Xu
  • , Ke Xu
  • , Furu Wei
  • *此作品的通讯作者
  • Stanford University
  • Microsoft USA
  • University of California at San Diego

科研成果: 书/报告/会议事项章节会议稿件同行评审

摘要

In this paper, we propose Sequence Span Rewriting (SSR), a self-supervised task for sequence-to-sequence (Seq2Seq) pre-training. SSR learns to refine the machine-generated imperfect text spans into ground truth text. SSR provides more fine-grained and informative supervision in addition to the original text-infilling objective. Compared to the prevalent text infilling objectives for Seq2Seq pretraining, SSR is naturally more consistent with many downstream generation tasks that require sentence rewriting (e.g., text summarization, question generation, grammatical error correction, and paraphrase generation). We conduct extensive experiments by using SSR to improve the typical Seq2Seq pre-trained model T5 in a continual pre-training setting and show substantial improvements over T5 on various natural language generation tasks.

源语言英语
主期刊名EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings
出版商Association for Computational Linguistics (ACL)
571-582
页数12
ISBN(电子版)9781955917094
DOI
出版状态已出版 - 2021
活动2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021 - Hybrid, Punta Cana, 多米尼加共和国
期限: 7 11月 202111 11月 2021

出版系列

姓名EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings

会议

会议2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021
国家/地区多米尼加共和国
Hybrid, Punta Cana
时期7/11/2111/11/21

指纹

探究 'Improving Sequence-to-Sequence Pre-training via Sequence Span Rewriting' 的科研主题。它们共同构成独一无二的指纹。

引用此