@inproceedings{imamura-sumita-2019-recycling,
title = "Recycling a Pre-trained {BERT} Encoder for Neural Machine Translation",
author = "Imamura, Kenji and
Sumita, Eiichiro",
editor = "Birch, Alexandra and
Finch, Andrew and
Hayashi, Hiroaki and
Konstas, Ioannis and
Luong, Thang and
Neubig, Graham and
Oda, Yusuke and
Sudoh, Katsuhito",
booktitle = "Proceedings of the 3rd Workshop on Neural Generation and Translation",
month = nov,
year = "2019",
address = "Hong Kong",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/D19-5603",
doi = "10.18653/v1/D19-5603",
pages = "23--31",
abstract = "In this paper, a pre-trained Bidirectional Encoder Representations from Transformers (BERT) model is applied to Transformer-based neural machine translation (NMT). In contrast to monolingual tasks, the number of unlearned model parameters in an NMT decoder is as huge as the number of learned parameters in the BERT model. To train all the models appropriately, we employ two-stage optimization, which first trains only the unlearned parameters by freezing the BERT model, and then fine-tunes all the sub-models. In our experiments, stable two-stage optimization was achieved, in contrast the BLEU scores of direct fine-tuning were extremely low. Consequently, the BLEU scores of the proposed method were better than those of the Transformer base model and the same model without pre-training. Additionally, we confirmed that NMT with the BERT encoder is more effective in low-resource settings.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="imamura-sumita-2019-recycling">
<titleInfo>
<title>Recycling a Pre-trained BERT Encoder for Neural Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kenji</namePart>
<namePart type="family">Imamura</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eiichiro</namePart>
<namePart type="family">Sumita</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2019-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 3rd Workshop on Neural Generation and Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Alexandra</namePart>
<namePart type="family">Birch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andrew</namePart>
<namePart type="family">Finch</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hiroaki</namePart>
<namePart type="family">Hayashi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ioannis</namePart>
<namePart type="family">Konstas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Thang</namePart>
<namePart type="family">Luong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Graham</namePart>
<namePart type="family">Neubig</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yusuke</namePart>
<namePart type="family">Oda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Katsuhito</namePart>
<namePart type="family">Sudoh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hong Kong</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>In this paper, a pre-trained Bidirectional Encoder Representations from Transformers (BERT) model is applied to Transformer-based neural machine translation (NMT). In contrast to monolingual tasks, the number of unlearned model parameters in an NMT decoder is as huge as the number of learned parameters in the BERT model. To train all the models appropriately, we employ two-stage optimization, which first trains only the unlearned parameters by freezing the BERT model, and then fine-tunes all the sub-models. In our experiments, stable two-stage optimization was achieved, in contrast the BLEU scores of direct fine-tuning were extremely low. Consequently, the BLEU scores of the proposed method were better than those of the Transformer base model and the same model without pre-training. Additionally, we confirmed that NMT with the BERT encoder is more effective in low-resource settings.</abstract>
<identifier type="citekey">imamura-sumita-2019-recycling</identifier>
<identifier type="doi">10.18653/v1/D19-5603</identifier>
<location>
<url>https://aclanthology.org/D19-5603</url>
</location>
<part>
<date>2019-11</date>
<extent unit="page">
<start>23</start>
<end>31</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Recycling a Pre-trained BERT Encoder for Neural Machine Translation
%A Imamura, Kenji
%A Sumita, Eiichiro
%Y Birch, Alexandra
%Y Finch, Andrew
%Y Hayashi, Hiroaki
%Y Konstas, Ioannis
%Y Luong, Thang
%Y Neubig, Graham
%Y Oda, Yusuke
%Y Sudoh, Katsuhito
%S Proceedings of the 3rd Workshop on Neural Generation and Translation
%D 2019
%8 November
%I Association for Computational Linguistics
%C Hong Kong
%F imamura-sumita-2019-recycling
%X In this paper, a pre-trained Bidirectional Encoder Representations from Transformers (BERT) model is applied to Transformer-based neural machine translation (NMT). In contrast to monolingual tasks, the number of unlearned model parameters in an NMT decoder is as huge as the number of learned parameters in the BERT model. To train all the models appropriately, we employ two-stage optimization, which first trains only the unlearned parameters by freezing the BERT model, and then fine-tunes all the sub-models. In our experiments, stable two-stage optimization was achieved, in contrast the BLEU scores of direct fine-tuning were extremely low. Consequently, the BLEU scores of the proposed method were better than those of the Transformer base model and the same model without pre-training. Additionally, we confirmed that NMT with the BERT encoder is more effective in low-resource settings.
%R 10.18653/v1/D19-5603
%U https://aclanthology.org/D19-5603
%U https://doi.org/10.18653/v1/D19-5603
%P 23-31
Markdown (Informal)
[Recycling a Pre-trained BERT Encoder for Neural Machine Translation](https://aclanthology.org/D19-5603) (Imamura & Sumita, NGT 2019)
ACL