@inproceedings{zheng-etal-2022-candidate,
title = "Candidate Soups: Fusing Candidate Results Improves Translation Quality for Non-Autoregressive Translation",
author = "Zheng, Huanran and
Zhu, Wei and
Wang, Pengfei and
Wang, Xiaoling",
editor = "Goldberg, Yoav and
Kozareva, Zornitsa and
Zhang, Yue",
booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.emnlp-main.318",
doi = "10.18653/v1/2022.emnlp-main.318",
pages = "4811--4823",
abstract = "Non-autoregressive translation (NAT) model achieves a much faster inference speed than the autoregressive translation (AT) model because it can simultaneously predict all tokens during inference. However, its translation quality suffers from degradation compared to AT. And existing NAT methods only focus on improving the NAT model{'}s performance but do not fully utilize it. In this paper, we propose a simple but effective method called {``}Candidate Soups,{''} which can obtain high-quality translations while maintaining the inference speed of NAT models. Unlike previous approaches that pick the individual result and discard the remainders, Candidate Soups (CDS) can fully use the valuable information in the different candidate translations through model uncertainty. Extensive experiments on two benchmarks (WMT{'}14 EN{--}DE and WMT{'}16 EN{--}RO) demonstrate the effectiveness and generality of our proposed method, which can significantly improve the translation quality of various base models. More notably, our best variant outperforms the AT model on three translation tasks with 7.6{\mbox{$\times$}} speedup.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zheng-etal-2022-candidate">
<titleInfo>
<title>Candidate Soups: Fusing Candidate Results Improves Translation Quality for Non-Autoregressive Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Huanran</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengfei</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoling</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yoav</namePart>
<namePart type="family">Goldberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zornitsa</namePart>
<namePart type="family">Kozareva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Non-autoregressive translation (NAT) model achieves a much faster inference speed than the autoregressive translation (AT) model because it can simultaneously predict all tokens during inference. However, its translation quality suffers from degradation compared to AT. And existing NAT methods only focus on improving the NAT model’s performance but do not fully utilize it. In this paper, we propose a simple but effective method called “Candidate Soups,” which can obtain high-quality translations while maintaining the inference speed of NAT models. Unlike previous approaches that pick the individual result and discard the remainders, Candidate Soups (CDS) can fully use the valuable information in the different candidate translations through model uncertainty. Extensive experiments on two benchmarks (WMT’14 EN–DE and WMT’16 EN–RO) demonstrate the effectiveness and generality of our proposed method, which can significantly improve the translation quality of various base models. More notably, our best variant outperforms the AT model on three translation tasks with 7.6\times speedup.</abstract>
<identifier type="citekey">zheng-etal-2022-candidate</identifier>
<identifier type="doi">10.18653/v1/2022.emnlp-main.318</identifier>
<location>
<url>https://aclanthology.org/2022.emnlp-main.318</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>4811</start>
<end>4823</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Candidate Soups: Fusing Candidate Results Improves Translation Quality for Non-Autoregressive Translation
%A Zheng, Huanran
%A Zhu, Wei
%A Wang, Pengfei
%A Wang, Xiaoling
%Y Goldberg, Yoav
%Y Kozareva, Zornitsa
%Y Zhang, Yue
%S Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates
%F zheng-etal-2022-candidate
%X Non-autoregressive translation (NAT) model achieves a much faster inference speed than the autoregressive translation (AT) model because it can simultaneously predict all tokens during inference. However, its translation quality suffers from degradation compared to AT. And existing NAT methods only focus on improving the NAT model’s performance but do not fully utilize it. In this paper, we propose a simple but effective method called “Candidate Soups,” which can obtain high-quality translations while maintaining the inference speed of NAT models. Unlike previous approaches that pick the individual result and discard the remainders, Candidate Soups (CDS) can fully use the valuable information in the different candidate translations through model uncertainty. Extensive experiments on two benchmarks (WMT’14 EN–DE and WMT’16 EN–RO) demonstrate the effectiveness and generality of our proposed method, which can significantly improve the translation quality of various base models. More notably, our best variant outperforms the AT model on three translation tasks with 7.6\times speedup.
%R 10.18653/v1/2022.emnlp-main.318
%U https://aclanthology.org/2022.emnlp-main.318
%U https://doi.org/10.18653/v1/2022.emnlp-main.318
%P 4811-4823
Markdown (Informal)
[Candidate Soups: Fusing Candidate Results Improves Translation Quality for Non-Autoregressive Translation](https://aclanthology.org/2022.emnlp-main.318) (Zheng et al., EMNLP 2022)
ACL