Large Language Models Robustness Against Perturbation

Large Language Models Robustness Against Perturbation | 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 Article Large Language Models Robustness Against Perturbation Saeed S. Alahmari, Lawerence Hall, Peter R. Mouton, Dmitry Goldgof This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7610884/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted 14 You are reading this latest preprint version Abstract Large Language Models (LLMs) have demonstrated impressive performance across various natural language processing (NLP) tasks, including text summarization, classification, and generation. Despite their success, LLMs are primarily trained on curated datasets that lack human-induced errors, such as typos or variations in word choice. As a result, LLMs may produce unexpected outputs when processing text containing such perturbations. In this paper, we investigate the resilience of LLMs to two types of text perturbations: typos and word substitutions. Using two public datasets, we evaluate the impact of these perturbations on text generation using six state-of-the-art models, including GPT-4o and LLaMA3.3-70B. Although previous studies have primarily examined the effects of perturbations in classification tasks, our research focuses on their impact on text generation. The results indicate that LLMs are sensitive to text perturbations, leading to variations in generated outputs, which have implications for their robustness and reliability in real-world applications. Biological sciences/Computational biology and bioinformatics Physical sciences/Mathematics and computing Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 29 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 06 Nov, 2025 Reviews received at journal 29 Oct, 2025 Reviewers agreed at journal 27 Oct, 2025 Reviews received at journal 24 Oct, 2025 Reviewers agreed at journal 24 Oct, 2025 Reviewers agreed at journal 24 Oct, 2025 Reviewers agreed at journal 21 Oct, 2025 Reviews received at journal 11 Oct, 2025 Reviewers agreed at journal 27 Sep, 2025 Reviewers invited by journal 27 Sep, 2025 Editor assigned by journal 22 Sep, 2025 Editor invited by journal 22 Sep, 2025 Submission checks completed at journal 18 Sep, 2025 First submitted to journal 18 Sep, 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-7610884","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":525839457,"identity":"16c007bc-c478-4899-80e0-2c64225e7205","order_by":0,"name":"Saeed S. Alahmari","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5ElEQVRIiWNgGAWjYHACxgOMDUCKmfkAmNdAjB6oFrYEUrUw8BgQp0W+//CDAz93bJM3b+f59piHwUZ2wwH2hx/waTG4kWZwsPfMbcM5h3m3G/MwpBlvOMBjLIFXiwSDwQHettuMM5h5t0nzMBxOBGphwKtFvv/4h4N/227bz2DmeQbU8h+ohf3xD7yeOZBjcBhoSyJQCxtQywGgFgYz/A67kVNwWLbtdvIMZjYzyTkGycYzD/OYWRBw2MaHb9tu287gP/xM4k2FnWzf8fbHN/A6DM1SIGYmQf0oGAWjYBSMAuwAAOo3TQGnwX60AAAAAElFTkSuQmCC","orcid":"","institution":"Najran University","correspondingAuthor":true,"prefix":"","firstName":"Saeed","middleName":"S.","lastName":"Alahmari","suffix":""},{"id":525839458,"identity":"d67e2cfe-865b-4903-ac4f-600807294a1c","order_by":1,"name":"Lawerence Hall","email":"","orcid":"","institution":"University of South Florida","correspondingAuthor":false,"prefix":"","firstName":"Lawerence","middleName":"","lastName":"Hall","suffix":""},{"id":525839460,"identity":"185a571b-ff84-41f8-befb-2500c96ff2d2","order_by":2,"name":"Peter R. Mouton","email":"","orcid":"","institution":"SRC Biosciences (United States)","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"R.","lastName":"Mouton","suffix":""},{"id":525839462,"identity":"1894b888-2171-4d42-962e-a4c97acfa0fc","order_by":3,"name":"Dmitry Goldgof","email":"","orcid":"","institution":"University of South Florida","correspondingAuthor":false,"prefix":"","firstName":"Dmitry","middleName":"","lastName":"Goldgof","suffix":""}],"badges":[],"createdAt":"2025-09-14 06:53:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7610884/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7610884/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-29770-0","type":"published","date":"2025-11-29T15:57:55+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":93100216,"identity":"e6f1011a-5757-48ac-9316-ecc4d9325579","added_by":"auto","created_at":"2025-10-09 04:57:42","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6145,"visible":true,"origin":"","legend":"","description":"","filename":"07622892a72a443ca5cdbc990188fb05.json","url":"https://assets-eu.researchsquare.com/files/rs-7610884/v1/95a8b32997655379c7f828bf.json"},{"id":97178408,"identity":"516d4061-6c0a-4360-b0fe-47855cbdf7dc","added_by":"auto","created_at":"2025-12-01 16:09:41","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":332158,"visible":true,"origin":"","legend":"","description":"","filename":"LargeLanguageModelsRobustnessAgainstPerturbation4.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7610884/v1_covered_24e3772d-c850-49c9-9f45-b64ff2613d6d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Large Language Models Robustness Against Perturbation","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7610884/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7610884/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Large Language Models (LLMs) have demonstrated impressive performance across various natural language processing (NLP) tasks, including text summarization, classification, and generation. Despite their success, LLMs are primarily trained on curated datasets that lack human-induced errors, such as typos or variations in word choice. As a result, LLMs may produce unexpected outputs when processing text containing such perturbations. In this paper, we investigate the resilience of LLMs to two types of text perturbations: typos and word substitutions. Using two public datasets, we evaluate the impact of these perturbations on text generation using six state-of-the-art models, including GPT-4o and LLaMA3.3-70B. Although previous studies have primarily examined the effects of perturbations in classification tasks, our research focuses on their impact on text generation. The results indicate that LLMs are sensitive to text perturbations, leading to variations in generated outputs, which have implications for their robustness and reliability in real-world applications.","manuscriptTitle":"Large Language Models Robustness Against Perturbation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-09 04:57:37","doi":"10.21203/rs.3.rs-7610884/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-06T14:32:56+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-29T14:47:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"321496697578869242285454349153138716203","date":"2025-10-27T11:58:25+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-24T22:09:15+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"286330639010873956187453288093168322055","date":"2025-10-24T22:02:33+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"152254546621071102528087854786190585805","date":"2025-10-24T07:02:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"30825447365456053188764209343105212503","date":"2025-10-22T00:01:16+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-11T12:45:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"277345414324961276301458910718888217234","date":"2025-09-27T09:26:36+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-27T08:43:47+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-22T09:14:07+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-22T08:50:32+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-18T05:17:55+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-09-18T04:57:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1c052042-98b0-4e56-9497-32e2c6aa1570","owner":[],"postedDate":"October 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":55882749,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":55882750,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2025-12-01T16:02:26+00:00","versionOfRecord":{"articleIdentity":"rs-7610884","link":"https://doi.org/10.1038/s41598-025-29770-0","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-11-29 15:57:55","publishedOnDateReadable":"November 29th, 2025"},"versionCreatedAt":"2025-10-09 04:57:37","video":"","vorDoi":"10.1038/s41598-025-29770-0","vorDoiUrl":"https://doi.org/10.1038/s41598-025-29770-0","workflowStages":[]},"version":"v1","identity":"rs-7610884","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7610884","identity":"rs-7610884","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}