GRAVITI: Grounded Retrieval Generation Framework for VideoLLM Hallucination Mitigation

GRAVITI: Grounded Retrieval Generation Framework for VideoLLM Hallucination Mitigation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article GRAVITI: Grounded Retrieval Generation Framework for VideoLLM Hallucination Mitigation Ahmad Khalil, Mahmoud Khalil, Alioune Ngom This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7900022/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 11 You are reading this latest preprint version Abstract Video-language models (VideoLLMs) excel at tasks such as video captioning and question answering, but often produce hallucinations—content not grounded in the video or metadata—limiting their reliability. To address this, we propose GRAVITI (Grounded Retrieval Generation framework for VideoLLM hallucination mitigation), a model-agnostic, training-free and API-free framework that integrates a dynamically constructed ad-hoc knowledge base with a retrieval-guided decoding process. We refer to this process as Grounded Retrieval Generation (GRG), where each generated token is conditioned on evidence retrieved from video features and auxiliary metadata. GRAVITI reduces hallucinations while remaining compatible across diverse VideoLLMs. Evaluated on three benchmarks—VidHalluc, EventHallusion, and VideoHallucer—GRAVITI improves overall accuracy by 6–14% and substantially lowers hallucination rates compared to strong baselines. Ablation studies demonstrate the impact of retrieval size, detector thresholds, and grounding mechanisms, highlighting the effectiveness of GRG in producing reliable, multi-modal video descriptions. VideoLLMs Hallucination GRAVITI Grounded Retrieval Generation Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 28 Jan, 2026 Reviews received at journal 23 Jan, 2026 Reviews received at journal 21 Jan, 2026 Reviewers agreed at journal 20 Jan, 2026 Reviewers agreed at journal 14 Jan, 2026 Reviews received at journal 11 Jan, 2026 Reviewers agreed at journal 15 Nov, 2025 Reviewers invited by journal 10 Nov, 2025 Editor assigned by journal 24 Oct, 2025 Submission checks completed at journal 24 Oct, 2025 First submitted to journal 19 Oct, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7900022","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":545393762,"identity":"800afa82-a786-4deb-ac7f-b49dc445d951","order_by":0,"name":"Ahmad Khalil","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3UlEQVRIie3QMQrCMBSA4UggLk9cX7F4hkhBHQSvUhHsUqGTOBVB0CsUvITe4EHASZw7Kh7Agks3TVyV2G4O+bcH+XhJGHO5/jBE1qAweUJbD8RENcIolNz3VrUIk3wkyYxViLdbX+giBQR53FPJIu0OVvxa2EjHP0p9MYB+HkuVnVXgkwjQRroYGoKazPeqtaFJxoD9IFGhiYQge5PUEF5aL4ax2RKCxDfhhgjrFi+LE00I8HQvFOi3IBf9oY1gHh2u5ZLG7e1s+gD9Y9hc33Ib+RKved7lcrlcn70A2D1HgK1BJqUAAAAASUVORK5CYII=","orcid":"","institution":"University of Windsor","correspondingAuthor":true,"prefix":"","firstName":"Ahmad","middleName":"","lastName":"Khalil","suffix":""},{"id":545393763,"identity":"3ecf4719-627b-4e56-8e34-913ba97bc333","order_by":1,"name":"Mahmoud Khalil","email":"","orcid":"","institution":"University of Windsor","correspondingAuthor":false,"prefix":"","firstName":"Mahmoud","middleName":"","lastName":"Khalil","suffix":""},{"id":545393765,"identity":"240773ab-4868-4769-af21-eb9e36e58adc","order_by":2,"name":"Alioune Ngom","email":"","orcid":"","institution":"University of Windsor","correspondingAuthor":false,"prefix":"","firstName":"Alioune","middleName":"","lastName":"Ngom","suffix":""}],"badges":[],"createdAt":"2025-10-19 17:23:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7900022/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7900022/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96316561,"identity":"998fcb9a-dc10-4cfe-853d-8636831bae79","added_by":"auto","created_at":"2025-11-19 17:47:21","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4383,"visible":true,"origin":"","legend":"","description":"","filename":"f9a406b98cae486d91b298ae5098e3f4.json","url":"https://assets-eu.researchsquare.com/files/rs-7900022/v1/18ddd185d5c398edb7ebe5bd.json"},{"id":96364736,"identity":"195a7139-f73d-47b2-98de-7b50f97b9bdd","added_by":"auto","created_at":"2025-11-20 10:09:35","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":410416,"visible":true,"origin":"","legend":"","description":"","filename":"jwpsxwnqnwdpjjtsxpvwyzrdpcccnwjw.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7900022/v1_covered_b0dcecc4-2683-4f75-a07b-f80ac8da1406.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"GRAVITI: Grounded Retrieval Generation Framework for VideoLLM Hallucination Mitigation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"VideoLLMs, Hallucination, GRAVITI, Grounded Retrieval Generation","lastPublishedDoi":"10.21203/rs.3.rs-7900022/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7900022/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Video-language models (VideoLLMs) excel at tasks such as video captioning and question answering, but often produce hallucinations—content not grounded in the video or metadata—limiting their reliability. To address this, we propose GRAVITI (Grounded Retrieval Generation framework for VideoLLM hallucination mitigation), a model-agnostic, training-free and API-free framework that integrates a dynamically constructed ad-hoc knowledge base with a retrieval-guided decoding process. We refer to this process as Grounded Retrieval Generation (GRG), where each generated token is conditioned on evidence retrieved from video features and auxiliary metadata. GRAVITI reduces hallucinations while remaining compatible across diverse VideoLLMs. Evaluated on three benchmarks—VidHalluc, EventHallusion, and VideoHallucer—GRAVITI improves overall accuracy by 6–14% and substantially lowers hallucination rates compared to strong baselines. Ablation studies demonstrate the impact of retrieval size, detector thresholds, and grounding mechanisms, highlighting the effectiveness of GRG in producing reliable, multi-modal video descriptions.","manuscriptTitle":"GRAVITI: Grounded Retrieval Generation Framework for VideoLLM Hallucination Mitigation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-19 17:47:16","doi":"10.21203/rs.3.rs-7900022/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-01-28T17:20:27+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-23T07:01:27+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-21T06:00:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"28361319553961326406891075047656843501","date":"2026-01-21T00:44:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"305766987312724715183608778301964910435","date":"2026-01-14T15:39:52+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-11T07:18:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"150237550768328713901851445576446474665","date":"2025-11-15T05:07:28+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-10T14:09:16+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-24T22:53:11+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-10-24T09:04:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"The Visual Computer","date":"2025-10-19T17:14:07+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"a063d82a-6adf-47cf-a906-7fd0bed66b98","owner":[],"postedDate":"November 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-01-28T17:24:28+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-19 17:47:16","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7900022","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7900022","identity":"rs-7900022","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}