@inproceedings{enouen-etal-2024-textgenshap,
title = "{T}ext{G}en{SHAP}: Scalable Post-Hoc Explanations in Text Generation with Long Documents",
author = "Enouen, James and
Nakhost, Hootan and
Ebrahimi, Sayna and
Arik, Sercan and
Liu, Yan and
Pfister, Tomas",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.832/",
doi = "10.18653/v1/2024.findings-acl.832",
pages = "13984--14011",
abstract = "Large language models (LLMs) have attracted great interest in many real-world applications; however, their ``black-box'' nature necessitates scalable and faithful explanations. Shapley values have matured as an explainability method for deep learning, but extending them to LLMs is difficult due to long input contexts and autoregressive output generation. We introduce , an efficient post-hoc explanation method incorporating LLM-specific techniques, which leads to significant runtime improvements: token-level explanations in minutes not hours, and document-level explanations within seconds. We demonstrate how such explanations can improve end-to-end performance of retrieval augmented generation by localizing important words within long documents and reranking passages collected by retrieval systems. On various open-domain question answering benchmarks, we show TextGenSHAP improves the retrieval recall and prediction accuracy significantly."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="enouen-etal-2024-textgenshap">
<titleInfo>
<title>TextGenSHAP: Scalable Post-Hoc Explanations in Text Generation with Long Documents</title>
</titleInfo>
<name type="personal">
<namePart type="given">James</namePart>
<namePart type="family">Enouen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hootan</namePart>
<namePart type="family">Nakhost</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sayna</namePart>
<namePart type="family">Ebrahimi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sercan</namePart>
<namePart type="family">Arik</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tomas</namePart>
<namePart type="family">Pfister</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2024</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lun-Wei</namePart>
<namePart type="family">Ku</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andre</namePart>
<namePart type="family">Martins</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Srikumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Large language models (LLMs) have attracted great interest in many real-world applications; however, their “black-box” nature necessitates scalable and faithful explanations. Shapley values have matured as an explainability method for deep learning, but extending them to LLMs is difficult due to long input contexts and autoregressive output generation. We introduce , an efficient post-hoc explanation method incorporating LLM-specific techniques, which leads to significant runtime improvements: token-level explanations in minutes not hours, and document-level explanations within seconds. We demonstrate how such explanations can improve end-to-end performance of retrieval augmented generation by localizing important words within long documents and reranking passages collected by retrieval systems. On various open-domain question answering benchmarks, we show TextGenSHAP improves the retrieval recall and prediction accuracy significantly.</abstract>
<identifier type="citekey">enouen-etal-2024-textgenshap</identifier>
<identifier type="doi">10.18653/v1/2024.findings-acl.832</identifier>
<location>
<url>https://aclanthology.org/2024.findings-acl.832/</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>13984</start>
<end>14011</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TextGenSHAP: Scalable Post-Hoc Explanations in Text Generation with Long Documents
%A Enouen, James
%A Nakhost, Hootan
%A Ebrahimi, Sayna
%A Arik, Sercan
%A Liu, Yan
%A Pfister, Tomas
%Y Ku, Lun-Wei
%Y Martins, Andre
%Y Srikumar, Vivek
%S Findings of the Association for Computational Linguistics: ACL 2024
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F enouen-etal-2024-textgenshap
%X Large language models (LLMs) have attracted great interest in many real-world applications; however, their “black-box” nature necessitates scalable and faithful explanations. Shapley values have matured as an explainability method for deep learning, but extending them to LLMs is difficult due to long input contexts and autoregressive output generation. We introduce , an efficient post-hoc explanation method incorporating LLM-specific techniques, which leads to significant runtime improvements: token-level explanations in minutes not hours, and document-level explanations within seconds. We demonstrate how such explanations can improve end-to-end performance of retrieval augmented generation by localizing important words within long documents and reranking passages collected by retrieval systems. On various open-domain question answering benchmarks, we show TextGenSHAP improves the retrieval recall and prediction accuracy significantly.
%R 10.18653/v1/2024.findings-acl.832
%U https://aclanthology.org/2024.findings-acl.832/
%U https://doi.org/10.18653/v1/2024.findings-acl.832
%P 13984-14011
Markdown (Informal)
[TextGenSHAP: Scalable Post-Hoc Explanations in Text Generation with Long Documents](https://aclanthology.org/2024.findings-acl.832/) (Enouen et al., Findings 2024)
ACL