@inproceedings{sinha-etal-2021-unnatural,
title = "{UnNatural} {L}anguage {I}nference",
author = "Sinha, Koustuv and
Parthasarathi, Prasanna and
Pineau, Joelle and
Williams, Adina",
editor = "Zong, Chengqing and
Xia, Fei and
Li, Wenjie and
Navigli, Roberto",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = aug,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-long.569/",
doi = "10.18653/v1/2021.acl-long.569",
pages = "7329--7346",
abstract = "Recent investigations into the inner-workings of state-of-the-art large-scale pre-trained Transformer-based Natural Language Understanding (NLU) models indicate that they appear to understand human-like syntax, at least to some extent. We provide novel evidence that complicates this claim: we find that state-of-the-art Natural Language Inference (NLI) models assign the same labels to permuted examples as they do to the original, i.e. they are invariant to random word-order permutations. This behavior notably differs from that of humans; we struggle to understand the meaning of ungrammatical sentences. To measure the severity of this issue, we propose a suite of metrics and investigate which properties of particular permutations lead models to be word order invariant. For example, in MNLI dataset we find almost all (98.7{\%}) examples contain at least one permutation which elicits the gold label. Models are even able to assign gold labels to permutations that they originally failed to predict correctly. We provide a comprehensive empirical evaluation of this phenomenon, and further show that this issue exists in pre-Transformer RNN / ConvNet based encoders, as well as across multiple languages (English and Chinese). Our code and data are available at \url{https://github.com/facebookresearch/unlu}."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sinha-etal-2021-unnatural">
<titleInfo>
<title>UnNatural Language Inference</title>
</titleInfo>
<name type="personal">
<namePart type="given">Koustuv</namePart>
<namePart type="family">Sinha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Prasanna</namePart>
<namePart type="family">Parthasarathi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joelle</namePart>
<namePart type="family">Pineau</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Adina</namePart>
<namePart type="family">Williams</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2021-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chengqing</namePart>
<namePart type="family">Zong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Xia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenjie</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Roberto</namePart>
<namePart type="family">Navigli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Recent investigations into the inner-workings of state-of-the-art large-scale pre-trained Transformer-based Natural Language Understanding (NLU) models indicate that they appear to understand human-like syntax, at least to some extent. We provide novel evidence that complicates this claim: we find that state-of-the-art Natural Language Inference (NLI) models assign the same labels to permuted examples as they do to the original, i.e. they are invariant to random word-order permutations. This behavior notably differs from that of humans; we struggle to understand the meaning of ungrammatical sentences. To measure the severity of this issue, we propose a suite of metrics and investigate which properties of particular permutations lead models to be word order invariant. For example, in MNLI dataset we find almost all (98.7%) examples contain at least one permutation which elicits the gold label. Models are even able to assign gold labels to permutations that they originally failed to predict correctly. We provide a comprehensive empirical evaluation of this phenomenon, and further show that this issue exists in pre-Transformer RNN / ConvNet based encoders, as well as across multiple languages (English and Chinese). Our code and data are available at https://github.com/facebookresearch/unlu.</abstract>
<identifier type="citekey">sinha-etal-2021-unnatural</identifier>
<identifier type="doi">10.18653/v1/2021.acl-long.569</identifier>
<location>
<url>https://aclanthology.org/2021.acl-long.569/</url>
</location>
<part>
<date>2021-08</date>
<extent unit="page">
<start>7329</start>
<end>7346</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T UnNatural Language Inference
%A Sinha, Koustuv
%A Parthasarathi, Prasanna
%A Pineau, Joelle
%A Williams, Adina
%Y Zong, Chengqing
%Y Xia, Fei
%Y Li, Wenjie
%Y Navigli, Roberto
%S Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
%D 2021
%8 August
%I Association for Computational Linguistics
%C Online
%F sinha-etal-2021-unnatural
%X Recent investigations into the inner-workings of state-of-the-art large-scale pre-trained Transformer-based Natural Language Understanding (NLU) models indicate that they appear to understand human-like syntax, at least to some extent. We provide novel evidence that complicates this claim: we find that state-of-the-art Natural Language Inference (NLI) models assign the same labels to permuted examples as they do to the original, i.e. they are invariant to random word-order permutations. This behavior notably differs from that of humans; we struggle to understand the meaning of ungrammatical sentences. To measure the severity of this issue, we propose a suite of metrics and investigate which properties of particular permutations lead models to be word order invariant. For example, in MNLI dataset we find almost all (98.7%) examples contain at least one permutation which elicits the gold label. Models are even able to assign gold labels to permutations that they originally failed to predict correctly. We provide a comprehensive empirical evaluation of this phenomenon, and further show that this issue exists in pre-Transformer RNN / ConvNet based encoders, as well as across multiple languages (English and Chinese). Our code and data are available at https://github.com/facebookresearch/unlu.
%R 10.18653/v1/2021.acl-long.569
%U https://aclanthology.org/2021.acl-long.569/
%U https://doi.org/10.18653/v1/2021.acl-long.569
%P 7329-7346
Markdown (Informal)
[UnNatural Language Inference](https://aclanthology.org/2021.acl-long.569/) (Sinha et al., ACL-IJCNLP 2021)
ACL
- Koustuv Sinha, Prasanna Parthasarathi, Joelle Pineau, and Adina Williams. 2021. UnNatural Language Inference. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 7329–7346, Online. Association for Computational Linguistics.