FastRM: An efficient and automatic explainability framework for multimodal generative models

While Large Vision Language Models (LVLMs) have become masterly capable in reasoning over human prompts and visual inputs, they are still prone to producing responses that contain misinformation. Identifying incorrect responses that are not grounded in evidence has become a crucial task in building...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Stan, Gabriela Ben-Melech, Aflalo, Estelle, Luo, Man, Rosenman, Shachar, Le, Tiep, Paul, Sayak, Tseng, Shao-Yen, Lal, Vasudev
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Stan, Gabriela Ben-Melech
Aflalo, Estelle
Luo, Man
Rosenman, Shachar
Le, Tiep
Paul, Sayak
Tseng, Shao-Yen
Lal, Vasudev
description While Large Vision Language Models (LVLMs) have become masterly capable in reasoning over human prompts and visual inputs, they are still prone to producing responses that contain misinformation. Identifying incorrect responses that are not grounded in evidence has become a crucial task in building trustworthy AI. Explainability methods such as gradient-based relevancy maps on LVLM outputs can provide an insight on the decision process of models, however these methods are often computationally expensive and not suited for on-the-fly validation of outputs. In this work, we propose FastRM, an effective way for predicting the explainable Relevancy Maps of LVLM models. Experimental results show that employing FastRM leads to a 99.8% reduction in compute time for relevancy map generation and an 44.4% reduction in memory footprint for the evaluated LVLM, making explainable AI more efficient and practical, thereby facilitating its deployment in real-world applications.
doi_str_mv 10.48550/arxiv.2412.01487
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2412_01487</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2412_01487</sourcerecordid><originalsourceid>FETCH-arxiv_primary_2412_014873</originalsourceid><addsrcrecordid>eNqFzsEKgkAUBdDZtIjqA1r1fiBTU5J2EUmbNtEykJe-iUczo4yj6d9n0r7Vhcu9cIRYBr4XJXHsb9B23HphFISeH0TJbiruKdbuetnDwQBJyTmTcYCmAGxcqdFxDtRVCtnggxW7HqRFTe_SvkCWFnSjHOuyQAVPMmSHR0swFKTquZhIVDUtfjkTq_R0O57XoyOrLGu0ffb1ZKNn-3_xAXdbQb4</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>FastRM: An efficient and automatic explainability framework for multimodal generative models</title><source>arXiv.org</source><creator>Stan, Gabriela Ben-Melech ; Aflalo, Estelle ; Luo, Man ; Rosenman, Shachar ; Le, Tiep ; Paul, Sayak ; Tseng, Shao-Yen ; Lal, Vasudev</creator><creatorcontrib>Stan, Gabriela Ben-Melech ; Aflalo, Estelle ; Luo, Man ; Rosenman, Shachar ; Le, Tiep ; Paul, Sayak ; Tseng, Shao-Yen ; Lal, Vasudev</creatorcontrib><description>While Large Vision Language Models (LVLMs) have become masterly capable in reasoning over human prompts and visual inputs, they are still prone to producing responses that contain misinformation. Identifying incorrect responses that are not grounded in evidence has become a crucial task in building trustworthy AI. Explainability methods such as gradient-based relevancy maps on LVLM outputs can provide an insight on the decision process of models, however these methods are often computationally expensive and not suited for on-the-fly validation of outputs. In this work, we propose FastRM, an effective way for predicting the explainable Relevancy Maps of LVLM models. Experimental results show that employing FastRM leads to a 99.8% reduction in compute time for relevancy map generation and an 44.4% reduction in memory footprint for the evaluated LVLM, making explainable AI more efficient and practical, thereby facilitating its deployment in real-world applications.</description><identifier>DOI: 10.48550/arxiv.2412.01487</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence</subject><creationdate>2024-12</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,777,882</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2412.01487$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2412.01487$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Stan, Gabriela Ben-Melech</creatorcontrib><creatorcontrib>Aflalo, Estelle</creatorcontrib><creatorcontrib>Luo, Man</creatorcontrib><creatorcontrib>Rosenman, Shachar</creatorcontrib><creatorcontrib>Le, Tiep</creatorcontrib><creatorcontrib>Paul, Sayak</creatorcontrib><creatorcontrib>Tseng, Shao-Yen</creatorcontrib><creatorcontrib>Lal, Vasudev</creatorcontrib><title>FastRM: An efficient and automatic explainability framework for multimodal generative models</title><description>While Large Vision Language Models (LVLMs) have become masterly capable in reasoning over human prompts and visual inputs, they are still prone to producing responses that contain misinformation. Identifying incorrect responses that are not grounded in evidence has become a crucial task in building trustworthy AI. Explainability methods such as gradient-based relevancy maps on LVLM outputs can provide an insight on the decision process of models, however these methods are often computationally expensive and not suited for on-the-fly validation of outputs. In this work, we propose FastRM, an effective way for predicting the explainable Relevancy Maps of LVLM models. Experimental results show that employing FastRM leads to a 99.8% reduction in compute time for relevancy map generation and an 44.4% reduction in memory footprint for the evaluated LVLM, making explainable AI more efficient and practical, thereby facilitating its deployment in real-world applications.</description><subject>Computer Science - Artificial Intelligence</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNqFzsEKgkAUBdDZtIjqA1r1fiBTU5J2EUmbNtEykJe-iUczo4yj6d9n0r7Vhcu9cIRYBr4XJXHsb9B23HphFISeH0TJbiruKdbuetnDwQBJyTmTcYCmAGxcqdFxDtRVCtnggxW7HqRFTe_SvkCWFnSjHOuyQAVPMmSHR0swFKTquZhIVDUtfjkTq_R0O57XoyOrLGu0ffb1ZKNn-3_xAXdbQb4</recordid><startdate>20241202</startdate><enddate>20241202</enddate><creator>Stan, Gabriela Ben-Melech</creator><creator>Aflalo, Estelle</creator><creator>Luo, Man</creator><creator>Rosenman, Shachar</creator><creator>Le, Tiep</creator><creator>Paul, Sayak</creator><creator>Tseng, Shao-Yen</creator><creator>Lal, Vasudev</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20241202</creationdate><title>FastRM: An efficient and automatic explainability framework for multimodal generative models</title><author>Stan, Gabriela Ben-Melech ; Aflalo, Estelle ; Luo, Man ; Rosenman, Shachar ; Le, Tiep ; Paul, Sayak ; Tseng, Shao-Yen ; Lal, Vasudev</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-arxiv_primary_2412_014873</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Computer Science - Artificial Intelligence</topic><toplevel>online_resources</toplevel><creatorcontrib>Stan, Gabriela Ben-Melech</creatorcontrib><creatorcontrib>Aflalo, Estelle</creatorcontrib><creatorcontrib>Luo, Man</creatorcontrib><creatorcontrib>Rosenman, Shachar</creatorcontrib><creatorcontrib>Le, Tiep</creatorcontrib><creatorcontrib>Paul, Sayak</creatorcontrib><creatorcontrib>Tseng, Shao-Yen</creatorcontrib><creatorcontrib>Lal, Vasudev</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Stan, Gabriela Ben-Melech</au><au>Aflalo, Estelle</au><au>Luo, Man</au><au>Rosenman, Shachar</au><au>Le, Tiep</au><au>Paul, Sayak</au><au>Tseng, Shao-Yen</au><au>Lal, Vasudev</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>FastRM: An efficient and automatic explainability framework for multimodal generative models</atitle><date>2024-12-02</date><risdate>2024</risdate><abstract>While Large Vision Language Models (LVLMs) have become masterly capable in reasoning over human prompts and visual inputs, they are still prone to producing responses that contain misinformation. Identifying incorrect responses that are not grounded in evidence has become a crucial task in building trustworthy AI. Explainability methods such as gradient-based relevancy maps on LVLM outputs can provide an insight on the decision process of models, however these methods are often computationally expensive and not suited for on-the-fly validation of outputs. In this work, we propose FastRM, an effective way for predicting the explainable Relevancy Maps of LVLM models. Experimental results show that employing FastRM leads to a 99.8% reduction in compute time for relevancy map generation and an 44.4% reduction in memory footprint for the evaluated LVLM, making explainable AI more efficient and practical, thereby facilitating its deployment in real-world applications.</abstract><doi>10.48550/arxiv.2412.01487</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2412.01487
ispartof
issn
language eng
recordid cdi_arxiv_primary_2412_01487
source arXiv.org
subjects Computer Science - Artificial Intelligence
title FastRM: An efficient and automatic explainability framework for multimodal generative models
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-20T05%3A17%3A35IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=FastRM:%20An%20efficient%20and%20automatic%20explainability%20framework%20for%20multimodal%20generative%20models&rft.au=Stan,%20Gabriela%20Ben-Melech&rft.date=2024-12-02&rft_id=info:doi/10.48550/arxiv.2412.01487&rft_dat=%3Carxiv_GOX%3E2412_01487%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true