@inproceedings{hagstrom-johansson-2022-adapt,
title = "How to Adapt Pre-trained Vision-and-Language Models to a Text-only Input?",
author = {Hagstr{\"o}m, Lovisa and
Johansson, Richard},
editor = "Calzolari, Nicoletta and
Huang, Chu-Ren and
Kim, Hansaem and
Pustejovsky, James and
Wanner, Leo and
Choi, Key-Sun and
Ryu, Pum-Mo and
Chen, Hsin-Hsi and
Donatelli, Lucia and
Ji, Heng and
Kurohashi, Sadao and
Paggio, Patrizia and
Xue, Nianwen and
Kim, Seokhwan and
Hahm, Younggyun and
He, Zhong and
Lee, Tony Kyungil and
Santus, Enrico and
Bond, Francis and
Na, Seung-Hoon",
booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
month = oct,
year = "2022",
address = "Gyeongju, Republic of Korea",
publisher = "International Committee on Computational Linguistics",
url = "https://aclanthology.org/2022.coling-1.494/",
pages = "5582--5596",
abstract = "Current language models have been criticised for learning language from text alone without connection between words and their meaning. Consequently, multimodal training has been proposed as a way for creating models with better language understanding by providing the lacking connection. We focus on pre-trained multimodal vision-and-language (VL) models for which there already are some results on their language understanding capabilities. An unresolved issue with evaluating the linguistic skills of these models, however, is that there is no established method for adapting them to text-only input without out-of-distribution uncertainty. To find the best approach, we investigate and compare seven possible methods for adapting three different pre-trained VL models to text-only input. Our evaluations on both GLUE and Visual Property Norms (VPN) show that care should be put into adapting VL models to zero-shot text-only tasks, while the models are less sensitive to how we adapt them to non-zero-shot tasks. We also find that the adaptation methods perform differently for different models and that unimodal model counterparts perform on par with the VL models regardless of adaptation, indicating that current VL models do not necessarily gain better language understanding from their multimodal training."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hagstrom-johansson-2022-adapt">
<titleInfo>
<title>How to Adapt Pre-trained Vision-and-Language Models to a Text-only Input?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lovisa</namePart>
<namePart type="family">Hagström</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Richard</namePart>
<namePart type="family">Johansson</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 29th International Conference on Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nicoletta</namePart>
<namePart type="family">Calzolari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chu-Ren</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hansaem</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">James</namePart>
<namePart type="family">Pustejovsky</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leo</namePart>
<namePart type="family">Wanner</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Key-Sun</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pum-Mo</namePart>
<namePart type="family">Ryu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hsin-Hsi</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lucia</namePart>
<namePart type="family">Donatelli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Heng</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sadao</namePart>
<namePart type="family">Kurohashi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Patrizia</namePart>
<namePart type="family">Paggio</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nianwen</namePart>
<namePart type="family">Xue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seokhwan</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Younggyun</namePart>
<namePart type="family">Hahm</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhong</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tony</namePart>
<namePart type="given">Kyungil</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Enrico</namePart>
<namePart type="family">Santus</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Francis</namePart>
<namePart type="family">Bond</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seung-Hoon</namePart>
<namePart type="family">Na</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>International Committee on Computational Linguistics</publisher>
<place>
<placeTerm type="text">Gyeongju, Republic of Korea</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Current language models have been criticised for learning language from text alone without connection between words and their meaning. Consequently, multimodal training has been proposed as a way for creating models with better language understanding by providing the lacking connection. We focus on pre-trained multimodal vision-and-language (VL) models for which there already are some results on their language understanding capabilities. An unresolved issue with evaluating the linguistic skills of these models, however, is that there is no established method for adapting them to text-only input without out-of-distribution uncertainty. To find the best approach, we investigate and compare seven possible methods for adapting three different pre-trained VL models to text-only input. Our evaluations on both GLUE and Visual Property Norms (VPN) show that care should be put into adapting VL models to zero-shot text-only tasks, while the models are less sensitive to how we adapt them to non-zero-shot tasks. We also find that the adaptation methods perform differently for different models and that unimodal model counterparts perform on par with the VL models regardless of adaptation, indicating that current VL models do not necessarily gain better language understanding from their multimodal training.</abstract>
<identifier type="citekey">hagstrom-johansson-2022-adapt</identifier>
<location>
<url>https://aclanthology.org/2022.coling-1.494/</url>
</location>
<part>
<date>2022-10</date>
<extent unit="page">
<start>5582</start>
<end>5596</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T How to Adapt Pre-trained Vision-and-Language Models to a Text-only Input?
%A Hagström, Lovisa
%A Johansson, Richard
%Y Calzolari, Nicoletta
%Y Huang, Chu-Ren
%Y Kim, Hansaem
%Y Pustejovsky, James
%Y Wanner, Leo
%Y Choi, Key-Sun
%Y Ryu, Pum-Mo
%Y Chen, Hsin-Hsi
%Y Donatelli, Lucia
%Y Ji, Heng
%Y Kurohashi, Sadao
%Y Paggio, Patrizia
%Y Xue, Nianwen
%Y Kim, Seokhwan
%Y Hahm, Younggyun
%Y He, Zhong
%Y Lee, Tony Kyungil
%Y Santus, Enrico
%Y Bond, Francis
%Y Na, Seung-Hoon
%S Proceedings of the 29th International Conference on Computational Linguistics
%D 2022
%8 October
%I International Committee on Computational Linguistics
%C Gyeongju, Republic of Korea
%F hagstrom-johansson-2022-adapt
%X Current language models have been criticised for learning language from text alone without connection between words and their meaning. Consequently, multimodal training has been proposed as a way for creating models with better language understanding by providing the lacking connection. We focus on pre-trained multimodal vision-and-language (VL) models for which there already are some results on their language understanding capabilities. An unresolved issue with evaluating the linguistic skills of these models, however, is that there is no established method for adapting them to text-only input without out-of-distribution uncertainty. To find the best approach, we investigate and compare seven possible methods for adapting three different pre-trained VL models to text-only input. Our evaluations on both GLUE and Visual Property Norms (VPN) show that care should be put into adapting VL models to zero-shot text-only tasks, while the models are less sensitive to how we adapt them to non-zero-shot tasks. We also find that the adaptation methods perform differently for different models and that unimodal model counterparts perform on par with the VL models regardless of adaptation, indicating that current VL models do not necessarily gain better language understanding from their multimodal training.
%U https://aclanthology.org/2022.coling-1.494/
%P 5582-5596
Markdown (Informal)
[How to Adapt Pre-trained Vision-and-Language Models to a Text-only Input?](https://aclanthology.org/2022.coling-1.494/) (Hagström & Johansson, COLING 2022)
ACL