Task-Dependent Sensitivity of VLA Models to Instruction Wording

Task-Dependent Sensitivity of VLA Models to Instruction Wording | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 31 March 2026 V1 Latest version Share on Task-Dependent Sensitivity of VLA Models to Instruction Wording Author : Jihwan Woo 0000-0002-0424-0242 [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.177497139.90709706/v1 116 views 74 downloads Contents Abstract Supplementary Material Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Vision-Language-Action (VLA) models condition robot actions on natural language, yet sensitivity to instruction wording has not been characterised. This letter evaluates OpenVLA-7B on three manipulation tasks, comparing action differences from synonymous rephrasing (e.g., ”put” vs. ”place” vs. ”set”) against differences from specificity variation (brief vs. step-by-step). Across 5 scenes per task with balanced comparisons (n = 50 pairs each), phrasing sensitivity is task-dependent: one task shows significantly larger phrasing than specificity differences (1.6x, p = 0.018), one shows no difference (p = 0.957), and one trends in the opposite direction (p = 0.092). In aggregate, phrasing and specificity produce comparable action differences (p = 0.395). Both exceed the stochastic noise floor by 2-4x. The results indicate that VLA instruction sensitivity is real but task-specific, and that deployment robustness cannot be assumed from single-task evaluation. Supplementary Material File (main.pdf) Download 332.16 KB Information & Authors Information Version history V1 Version 1 31 March 2026 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords artificial intelligence robots Authors Affiliations Jihwan Woo 0000-0002-0424-0242 [email protected] Amazon Web Services Inc View all articles by this author Metrics & Citations Metrics Article Usage 116 views 74 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Jihwan Woo. Task-Dependent Sensitivity of VLA Models to Instruction Wording. Authorea . 31 March 2026. DOI: https://doi.org/10.22541/au.177497139.90709706/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.177497139.90709706/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe5a2b1ab10ad07',t:'MTc3OTIyMDI3OA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();