<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="marecek-rosa-2019-balustrades">
    <titleInfo>
        <title>From Balustrades to Pierre Vinken: Looking for Syntax in Transformer Self-Attentions</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">David</namePart>
        <namePart type="family">Mareček</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Rudolf</namePart>
        <namePart type="family">Rosa</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2019-08</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Tal</namePart>
            <namePart type="family">Linzen</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Grzegorz</namePart>
            <namePart type="family">Chrupała</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Yonatan</namePart>
            <namePart type="family">Belinkov</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Dieuwke</namePart>
            <namePart type="family">Hupkes</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Florence, Italy</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>We inspect the multi-head self-attention in Transformer NMT encoders for three source languages, looking for patterns that could have a syntactic interpretation. In many of the attention heads, we frequently find sequences of consecutive states attending to the same position, which resemble syntactic phrases. We propose a transparent deterministic method of quantifying the amount of syntactic information present in the self-attentions, based on automatically building and evaluating phrase-structure trees from the phrase-like sequences. We compare the resulting trees to existing constituency treebanks, both manually and by computing precision and recall.</abstract>
    <identifier type="citekey">marecek-rosa-2019-balustrades</identifier>
    <identifier type="doi">10.18653/v1/W19-4827</identifier>
    <location>
        <url>https://aclanthology.org/W19-4827/</url>
    </location>
    <part>
        <date>2019-08</date>
        <extent unit="page">
            <start>263</start>
            <end>275</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T From Balustrades to Pierre Vinken: Looking for Syntax in Transformer Self-Attentions
%A Mareček, David
%A Rosa, Rudolf
%Y Linzen, Tal
%Y Chrupała, Grzegorz
%Y Belinkov, Yonatan
%Y Hupkes, Dieuwke
%S Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP
%D 2019
%8 August
%I Association for Computational Linguistics
%C Florence, Italy
%F marecek-rosa-2019-balustrades
%X We inspect the multi-head self-attention in Transformer NMT encoders for three source languages, looking for patterns that could have a syntactic interpretation. In many of the attention heads, we frequently find sequences of consecutive states attending to the same position, which resemble syntactic phrases. We propose a transparent deterministic method of quantifying the amount of syntactic information present in the self-attentions, based on automatically building and evaluating phrase-structure trees from the phrase-like sequences. We compare the resulting trees to existing constituency treebanks, both manually and by computing precision and recall.
%R 10.18653/v1/W19-4827
%U https://aclanthology.org/W19-4827/
%U https://doi.org/10.18653/v1/W19-4827
%P 263-275

Markdown (Informal)

[From Balustrades to Pierre Vinken: Looking for Syntax in Transformer Self-Attentions](https://aclanthology.org/W19-4827/) (Mareček & Rosa, BlackboxNLP 2019)

• From Balustrades to Pierre Vinken: Looking for Syntax in Transformer Self-Attentions (Mareček & Rosa, BlackboxNLP 2019)

ACL

• David Mareček and Rudolf Rosa. 2019. From Balustrades to Pierre Vinken: Looking for Syntax in Transformer Self-Attentions. In Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pages 263–275, Florence, Italy. Association for Computational Linguistics.