@inproceedings{gowda-etal-2023-cometoid,
title = "Cometoid: Distilling Strong Reference-based Machine Translation Metrics into {E}ven Stronger Quality Estimation Metrics",
author = "Gowda, Thamme and
Kocmi, Tom and
Junczys-Dowmunt, Marcin",
editor = "Koehn, Philipp and
Haddow, Barry and
Kocmi, Tom and
Monz, Christof",
booktitle = "Proceedings of the Eighth Conference on Machine Translation",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.wmt-1.62",
doi = "10.18653/v1/2023.wmt-1.62",
pages = "751--755",
abstract = "This paper describes our submissions to the 2023 Conference on Machine Translation (WMT-23) Metrics shared task. Knowledge distillation is commonly used to create smaller student models that mimic larger teacher model while reducing the model size and hence inference cost in production. In this work, we apply knowledge distillation to machine translation evaluation metrics and distill existing reference-based teacher metrics into reference-free (quality estimation; QE) student metrics. We mainly focus on students of Unbabel{'}s COMET22 reference-based metric. When evaluating on the official WMT-22 Metrics evaluation task, our distilled Cometoid QE metrics outperform all other QE metrics on that set while matching or out-performing the reference-based teacher metric. Our metrics never see the human ground-truth scores directly {--} only the teacher metric was trained on human scores by its original creators. We also distill ChrF sentence-level scores into a neural QE metric and find that our reference-free (and fully human-score-free) student metric ChrFoid outperforms its teacher metric by over 7{\%} pairwise accuracy on the same WMT-22 task, rivaling other existing QE metrics.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gowda-etal-2023-cometoid">
<titleInfo>
<title>Cometoid: Distilling Strong Reference-based Machine Translation Metrics into Even Stronger Quality Estimation Metrics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Thamme</namePart>
<namePart type="family">Gowda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tom</namePart>
<namePart type="family">Kocmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marcin</namePart>
<namePart type="family">Junczys-Dowmunt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Eighth Conference on Machine Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Philipp</namePart>
<namePart type="family">Koehn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Barry</namePart>
<namePart type="family">Haddow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tom</namePart>
<namePart type="family">Kocmi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christof</namePart>
<namePart type="family">Monz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Singapore</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper describes our submissions to the 2023 Conference on Machine Translation (WMT-23) Metrics shared task. Knowledge distillation is commonly used to create smaller student models that mimic larger teacher model while reducing the model size and hence inference cost in production. In this work, we apply knowledge distillation to machine translation evaluation metrics and distill existing reference-based teacher metrics into reference-free (quality estimation; QE) student metrics. We mainly focus on students of Unbabel’s COMET22 reference-based metric. When evaluating on the official WMT-22 Metrics evaluation task, our distilled Cometoid QE metrics outperform all other QE metrics on that set while matching or out-performing the reference-based teacher metric. Our metrics never see the human ground-truth scores directly – only the teacher metric was trained on human scores by its original creators. We also distill ChrF sentence-level scores into a neural QE metric and find that our reference-free (and fully human-score-free) student metric ChrFoid outperforms its teacher metric by over 7% pairwise accuracy on the same WMT-22 task, rivaling other existing QE metrics.</abstract>
<identifier type="citekey">gowda-etal-2023-cometoid</identifier>
<identifier type="doi">10.18653/v1/2023.wmt-1.62</identifier>
<location>
<url>https://aclanthology.org/2023.wmt-1.62</url>
</location>
<part>
<date>2023-12</date>
<extent unit="page">
<start>751</start>
<end>755</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Cometoid: Distilling Strong Reference-based Machine Translation Metrics into Even Stronger Quality Estimation Metrics
%A Gowda, Thamme
%A Kocmi, Tom
%A Junczys-Dowmunt, Marcin
%Y Koehn, Philipp
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Monz, Christof
%S Proceedings of the Eighth Conference on Machine Translation
%D 2023
%8 December
%I Association for Computational Linguistics
%C Singapore
%F gowda-etal-2023-cometoid
%X This paper describes our submissions to the 2023 Conference on Machine Translation (WMT-23) Metrics shared task. Knowledge distillation is commonly used to create smaller student models that mimic larger teacher model while reducing the model size and hence inference cost in production. In this work, we apply knowledge distillation to machine translation evaluation metrics and distill existing reference-based teacher metrics into reference-free (quality estimation; QE) student metrics. We mainly focus on students of Unbabel’s COMET22 reference-based metric. When evaluating on the official WMT-22 Metrics evaluation task, our distilled Cometoid QE metrics outperform all other QE metrics on that set while matching or out-performing the reference-based teacher metric. Our metrics never see the human ground-truth scores directly – only the teacher metric was trained on human scores by its original creators. We also distill ChrF sentence-level scores into a neural QE metric and find that our reference-free (and fully human-score-free) student metric ChrFoid outperforms its teacher metric by over 7% pairwise accuracy on the same WMT-22 task, rivaling other existing QE metrics.
%R 10.18653/v1/2023.wmt-1.62
%U https://aclanthology.org/2023.wmt-1.62
%U https://doi.org/10.18653/v1/2023.wmt-1.62
%P 751-755
Markdown (Informal)
[Cometoid: Distilling Strong Reference-based Machine Translation Metrics into Even Stronger Quality Estimation Metrics](https://aclanthology.org/2023.wmt-1.62) (Gowda et al., WMT 2023)
ACL