@inproceedings{petrak-etal-2023-arithmetic,
title = "Arithmetic-Based Pretraining Improving Numeracy of Pretrained Language Models",
author = "Petrak, Dominic and
Moosavi, Nafise Sadat and
Gurevych, Iryna",
editor = "Palmer, Alexis and
Camacho-collados, Jose",
booktitle = "Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM 2023)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.starsem-1.42",
doi = "10.18653/v1/2023.starsem-1.42",
pages = "477--493",
abstract = "State-of-the-art pretrained language models tend to perform below their capabilities when applied out-of-the-box on tasks that require understanding and working with numbers (usually referred to as numeracy). Recent work suggests two main reasons for this: (1) popular tokenisation algorithms have limited expressiveness for numbers, and (2) common pretraining objectives do not target numeracy. Approaches that address these shortcomings usually require architectural changes or pretraining from scratch. In this paper, we propose a new extended pretraining approach called Arithmetic-Based Pretraining that jointly addresses both in one extended pretraining step without requiring architectural changes or pretraining from scratch. Arithmetic-Based Pretraining combines contrastive learning to improve the number representation, and a novel extended pretraining objective called Inferable Number Prediction Task to improve numeracy. Our experiments show the effectiveness of Arithmetic-Based Pretraining in three different tasks that require improved numeracy, i.e., reading comprehension in the DROP dataset, inference-on-tables in the InfoTabs dataset, and table-to-text generation in the WikiBio and SciGen datasets.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="petrak-etal-2023-arithmetic">
<titleInfo>
<title>Arithmetic-Based Pretraining Improving Numeracy of Pretrained Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dominic</namePart>
<namePart type="family">Petrak</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nafise</namePart>
<namePart type="given">Sadat</namePart>
<namePart type="family">Moosavi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Iryna</namePart>
<namePart type="family">Gurevych</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>Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM 2023)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Alexis</namePart>
<namePart type="family">Palmer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jose</namePart>
<namePart type="family">Camacho-collados</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>State-of-the-art pretrained language models tend to perform below their capabilities when applied out-of-the-box on tasks that require understanding and working with numbers (usually referred to as numeracy). Recent work suggests two main reasons for this: (1) popular tokenisation algorithms have limited expressiveness for numbers, and (2) common pretraining objectives do not target numeracy. Approaches that address these shortcomings usually require architectural changes or pretraining from scratch. In this paper, we propose a new extended pretraining approach called Arithmetic-Based Pretraining that jointly addresses both in one extended pretraining step without requiring architectural changes or pretraining from scratch. Arithmetic-Based Pretraining combines contrastive learning to improve the number representation, and a novel extended pretraining objective called Inferable Number Prediction Task to improve numeracy. Our experiments show the effectiveness of Arithmetic-Based Pretraining in three different tasks that require improved numeracy, i.e., reading comprehension in the DROP dataset, inference-on-tables in the InfoTabs dataset, and table-to-text generation in the WikiBio and SciGen datasets.</abstract>
<identifier type="citekey">petrak-etal-2023-arithmetic</identifier>
<identifier type="doi">10.18653/v1/2023.starsem-1.42</identifier>
<location>
<url>https://aclanthology.org/2023.starsem-1.42</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>477</start>
<end>493</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Arithmetic-Based Pretraining Improving Numeracy of Pretrained Language Models
%A Petrak, Dominic
%A Moosavi, Nafise Sadat
%A Gurevych, Iryna
%Y Palmer, Alexis
%Y Camacho-collados, Jose
%S Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM 2023)
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F petrak-etal-2023-arithmetic
%X State-of-the-art pretrained language models tend to perform below their capabilities when applied out-of-the-box on tasks that require understanding and working with numbers (usually referred to as numeracy). Recent work suggests two main reasons for this: (1) popular tokenisation algorithms have limited expressiveness for numbers, and (2) common pretraining objectives do not target numeracy. Approaches that address these shortcomings usually require architectural changes or pretraining from scratch. In this paper, we propose a new extended pretraining approach called Arithmetic-Based Pretraining that jointly addresses both in one extended pretraining step without requiring architectural changes or pretraining from scratch. Arithmetic-Based Pretraining combines contrastive learning to improve the number representation, and a novel extended pretraining objective called Inferable Number Prediction Task to improve numeracy. Our experiments show the effectiveness of Arithmetic-Based Pretraining in three different tasks that require improved numeracy, i.e., reading comprehension in the DROP dataset, inference-on-tables in the InfoTabs dataset, and table-to-text generation in the WikiBio and SciGen datasets.
%R 10.18653/v1/2023.starsem-1.42
%U https://aclanthology.org/2023.starsem-1.42
%U https://doi.org/10.18653/v1/2023.starsem-1.42
%P 477-493
Markdown (Informal)
[Arithmetic-Based Pretraining Improving Numeracy of Pretrained Language Models](https://aclanthology.org/2023.starsem-1.42) (Petrak et al., *SEM 2023)
ACL