@inproceedings{michaelov-bergen-2023-rarely,
title = "Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers",
author = "Michaelov, James and
Bergen, Benjamin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-acl.891",
doi = "10.18653/v1/2023.findings-acl.891",
pages = "14162--14174",
abstract = "How well do language models deal with quantification? In this study, we focus on {`}few{'}-type quantifiers, as in {`}few children like toys{'}, which might pose a particular challenge for language models because the sentence components with out the quantifier are likely to co-occur, and {`}few{'}-type quantifiers are rare. We present 960 English sentence stimuli from two human neurolinguistic experiments to 22 autoregressive transformer models of differing sizes. Not only do all the models perform poorly on {`}few{'}-type quantifiers, but overall the larger the model, the worse its performance. This inverse scaling is consistent with previous work suggesting that larger models increasingly reflect online rather than offline human processing, and we argue that the decreasing performance of larger models may challenge uses of language models as the basis for natural language systems.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="michaelov-bergen-2023-rarely">
<titleInfo>
<title>Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers</title>
</titleInfo>
<name type="personal">
<namePart type="given">James</namePart>
<namePart type="family">Michaelov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Benjamin</namePart>
<namePart type="family">Bergen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2023</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anna</namePart>
<namePart type="family">Rogers</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jordan</namePart>
<namePart type="family">Boyd-Graber</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Naoaki</namePart>
<namePart type="family">Okazaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>How well do language models deal with quantification? In this study, we focus on ‘few’-type quantifiers, as in ‘few children like toys’, which might pose a particular challenge for language models because the sentence components with out the quantifier are likely to co-occur, and ‘few’-type quantifiers are rare. We present 960 English sentence stimuli from two human neurolinguistic experiments to 22 autoregressive transformer models of differing sizes. Not only do all the models perform poorly on ‘few’-type quantifiers, but overall the larger the model, the worse its performance. This inverse scaling is consistent with previous work suggesting that larger models increasingly reflect online rather than offline human processing, and we argue that the decreasing performance of larger models may challenge uses of language models as the basis for natural language systems.</abstract>
<identifier type="citekey">michaelov-bergen-2023-rarely</identifier>
<identifier type="doi">10.18653/v1/2023.findings-acl.891</identifier>
<location>
<url>https://aclanthology.org/2023.findings-acl.891</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>14162</start>
<end>14174</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers
%A Michaelov, James
%A Bergen, Benjamin
%Y Rogers, Anna
%Y Boyd-Graber, Jordan
%Y Okazaki, Naoaki
%S Findings of the Association for Computational Linguistics: ACL 2023
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F michaelov-bergen-2023-rarely
%X How well do language models deal with quantification? In this study, we focus on ‘few’-type quantifiers, as in ‘few children like toys’, which might pose a particular challenge for language models because the sentence components with out the quantifier are likely to co-occur, and ‘few’-type quantifiers are rare. We present 960 English sentence stimuli from two human neurolinguistic experiments to 22 autoregressive transformer models of differing sizes. Not only do all the models perform poorly on ‘few’-type quantifiers, but overall the larger the model, the worse its performance. This inverse scaling is consistent with previous work suggesting that larger models increasingly reflect online rather than offline human processing, and we argue that the decreasing performance of larger models may challenge uses of language models as the basis for natural language systems.
%R 10.18653/v1/2023.findings-acl.891
%U https://aclanthology.org/2023.findings-acl.891
%U https://doi.org/10.18653/v1/2023.findings-acl.891
%P 14162-14174
Markdown (Informal)
[Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers](https://aclanthology.org/2023.findings-acl.891) (Michaelov & Bergen, Findings 2023)
ACL