Train Once, and Decode As You Like

Chao Tian, Yifei Wang, Hao Cheng, Yijiang Lian, Zhihua Zhang


Abstract
In this paper we propose a unified approach for supporting different generation manners of machine translation, including autoregressive, semi-autoregressive, and refinement-based non-autoregressive models. Our approach works by repeatedly selecting positions and generating tokens at these selected positions. After being trained once, our approach achieves better or competitive translation performance compared with some strong task-specific baseline models in all the settings. This generalization ability benefits mainly from the new training objective that we propose. We validate our approach on the WMT’14 English-German and IWSLT’14 German-English translation tasks. The experimental results are encouraging.
Anthology ID:
2020.coling-main.25
Volume:
Proceedings of the 28th International Conference on Computational Linguistics
Month:
December
Year:
2020
Address:
Barcelona, Spain (Online)
Editors:
Donia Scott, Nuria Bel, Chengqing Zong
Venue:
COLING
SIG:
Publisher:
International Committee on Computational Linguistics
Note:
Pages:
280–293
Language:
URL:
https://aclanthology.org/2020.coling-main.25
DOI:
10.18653/v1/2020.coling-main.25
Bibkey:
Cite (ACL):
Chao Tian, Yifei Wang, Hao Cheng, Yijiang Lian, and Zhihua Zhang. 2020. Train Once, and Decode As You Like. In Proceedings of the 28th International Conference on Computational Linguistics, pages 280–293, Barcelona, Spain (Online). International Committee on Computational Linguistics.
Cite (Informal):
Train Once, and Decode As You Like (Tian et al., COLING 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.coling-main.25.pdf