<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="song-etal-2024-better">
    <titleInfo>
        <title>Better Explain Transformers by Illuminating Important Information</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Linxin</namePart>
        <namePart type="family">Song</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yan</namePart>
        <namePart type="family">Cui</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ao</namePart>
        <namePart type="family">Luo</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Freddy</namePart>
        <namePart type="family">Lecue</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Irene</namePart>
        <namePart type="family">Li</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2024-03</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: EACL 2024</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Yvette</namePart>
            <namePart type="family">Graham</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Matthew</namePart>
            <namePart type="family">Purver</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">St. Julian’s, Malta</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Transformer-based models excel in various natural language processing (NLP) tasks, attracting countless efforts to explain their inner workings. Prior methods explain Transformers by focusing on the raw gradient and attention as token attribution scores, where non-relevant information is often considered during explanation computation, resulting in confusing results. In this work, we propose highlighting the important information and eliminating irrelevant information by a refined information flow on top of the layer-wise relevance propagation (LRP) method. Specifically, we consider identifying syntactic and positional heads as important attention heads and focus on the relevance obtained from these important heads. Experimental results demonstrate that irrelevant information does distort output attribution scores and then should be masked during explanation computation. Compared to eight baselines on both classification and question-answering datasets, our method consistently outperforms with over 3% to 33% improvement on explanation metrics, providing superior explanation performance. Our anonymous code repository is available at: https://anonymous.4open.science/r/MLRP-E676/</abstract>
    <identifier type="citekey">song-etal-2024-better</identifier>
    <location>
        <url>https://aclanthology.org/2024.findings-eacl.138</url>
    </location>
    <part>
        <date>2024-03</date>
        <extent unit="page">
            <start>2048</start>
            <end>2062</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Better Explain Transformers by Illuminating Important Information
%A Song, Linxin
%A Cui, Yan
%A Luo, Ao
%A Lecue, Freddy
%A Li, Irene
%Y Graham, Yvette
%Y Purver, Matthew
%S Findings of the Association for Computational Linguistics: EACL 2024
%D 2024
%8 March
%I Association for Computational Linguistics
%C St. Julian’s, Malta
%F song-etal-2024-better
%X Transformer-based models excel in various natural language processing (NLP) tasks, attracting countless efforts to explain their inner workings. Prior methods explain Transformers by focusing on the raw gradient and attention as token attribution scores, where non-relevant information is often considered during explanation computation, resulting in confusing results. In this work, we propose highlighting the important information and eliminating irrelevant information by a refined information flow on top of the layer-wise relevance propagation (LRP) method. Specifically, we consider identifying syntactic and positional heads as important attention heads and focus on the relevance obtained from these important heads. Experimental results demonstrate that irrelevant information does distort output attribution scores and then should be masked during explanation computation. Compared to eight baselines on both classification and question-answering datasets, our method consistently outperforms with over 3% to 33% improvement on explanation metrics, providing superior explanation performance. Our anonymous code repository is available at: https://anonymous.4open.science/r/MLRP-E676/
%U https://aclanthology.org/2024.findings-eacl.138
%P 2048-2062

Markdown (Informal)

[Better Explain Transformers by Illuminating Important Information](https://aclanthology.org/2024.findings-eacl.138) (Song et al., Findings 2024)

• Better Explain Transformers by Illuminating Important Information (Song et al., Findings 2024)

ACL

• Linxin Song, Yan Cui, Ao Luo, Freddy Lecue, and Irene Li. 2024. Better Explain Transformers by Illuminating Important Information. In Findings of the Association for Computational Linguistics: EACL 2024, pages 2048–2062, St. Julian’s, Malta. Association for Computational Linguistics.