@inproceedings{sia-duh-2022-prefix,
title = "Prefix Embeddings for In-context Machine Translation",
author = "Sia, Suzanna and
Duh, Kevin",
editor = "Duh, Kevin and
Guzm{\'a}n, Francisco",
booktitle = "Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track)",
month = sep,
year = "2022",
address = "Orlando, USA",
publisher = "Association for Machine Translation in the Americas",
url = "https://aclanthology.org/2022.amta-research.4",
pages = "45--57",
abstract = "Very large language models have been shown to translate with few-shot in-context examples. However, they have not achieved state-of-art results for translating out of English. In this work, we investigate an extremely lightweight fixed-parameter method for conditioning a large language model to better translate into the target language. Our method introduces additional embeddings, known as prefix embeddings which do not interfere with the existing weights of the model. Using unsupervised and weakly semi-supervised methods that train only 0.0001{\%} of the model parameters, the simple method improves {\textasciitilde}0.2-1.3 BLEU points across 3 domains and 3 languages. We analyze the resulting embeddings{'} training dynamics, and where they lie in the embedding space, and show that our trained embeddings can be used for both in-context translation, and diverse generation of the target sentence.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sia-duh-2022-prefix">
<titleInfo>
<title>Prefix Embeddings for In-context Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Suzanna</namePart>
<namePart type="family">Sia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kevin</namePart>
<namePart type="family">Duh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-09</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kevin</namePart>
<namePart type="family">Duh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Francisco</namePart>
<namePart type="family">Guzmán</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Machine Translation in the Americas</publisher>
<place>
<placeTerm type="text">Orlando, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Very large language models have been shown to translate with few-shot in-context examples. However, they have not achieved state-of-art results for translating out of English. In this work, we investigate an extremely lightweight fixed-parameter method for conditioning a large language model to better translate into the target language. Our method introduces additional embeddings, known as prefix embeddings which do not interfere with the existing weights of the model. Using unsupervised and weakly semi-supervised methods that train only 0.0001% of the model parameters, the simple method improves ~0.2-1.3 BLEU points across 3 domains and 3 languages. We analyze the resulting embeddings’ training dynamics, and where they lie in the embedding space, and show that our trained embeddings can be used for both in-context translation, and diverse generation of the target sentence.</abstract>
<identifier type="citekey">sia-duh-2022-prefix</identifier>
<location>
<url>https://aclanthology.org/2022.amta-research.4</url>
</location>
<part>
<date>2022-09</date>
<extent unit="page">
<start>45</start>
<end>57</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Prefix Embeddings for In-context Machine Translation
%A Sia, Suzanna
%A Duh, Kevin
%Y Duh, Kevin
%Y Guzmán, Francisco
%S Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track)
%D 2022
%8 September
%I Association for Machine Translation in the Americas
%C Orlando, USA
%F sia-duh-2022-prefix
%X Very large language models have been shown to translate with few-shot in-context examples. However, they have not achieved state-of-art results for translating out of English. In this work, we investigate an extremely lightweight fixed-parameter method for conditioning a large language model to better translate into the target language. Our method introduces additional embeddings, known as prefix embeddings which do not interfere with the existing weights of the model. Using unsupervised and weakly semi-supervised methods that train only 0.0001% of the model parameters, the simple method improves ~0.2-1.3 BLEU points across 3 domains and 3 languages. We analyze the resulting embeddings’ training dynamics, and where they lie in the embedding space, and show that our trained embeddings can be used for both in-context translation, and diverse generation of the target sentence.
%U https://aclanthology.org/2022.amta-research.4
%P 45-57
Markdown (Informal)
[Prefix Embeddings for In-context Machine Translation](https://aclanthology.org/2022.amta-research.4) (Sia & Duh, AMTA 2022)
ACL
- Suzanna Sia and Kevin Duh. 2022. Prefix Embeddings for In-context Machine Translation. In Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track), pages 45–57, Orlando, USA. Association for Machine Translation in the Americas.