{"id":6168,"date":"2026-10-19T16:49:43","date_gmt":"2026-10-19T16:49:43","guid":{"rendered":"https:\/\/i10xnews.kinsta.cloud\/?p=6168"},"modified":"2026-10-19T17:24:20","modified_gmt":"2026-10-19T17:24:20","slug":"stuck-on-shuffle-attribute-object-binding-vlm","status":"publish","type":"post","link":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm","title":{"rendered":"VLMs Still Can&#8217;t Tell Which Attribute Goes With Which Object \u2014 Until Text-Only Finetuning"},"content":{"rendered":"\n<h2 class=\"wp-block-heading\">\u26a1 Quick Take<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>The finding:<\/strong> Attribute\u2013object binding remains a weak spot: models confuse which adjective belongs to which noun across modalities.<\/li><li><strong>Text-only fix:<\/strong> Finetuning an 8B model on synthetic shuffled-caption supervision (no images) raises mean accuracy from 52.5% to 84.2%, with 99.3% text\u2013text accuracy.<\/li><li><strong>Cross-modal transfer:<\/strong> Despite no visual data, gains of 34.1 \/ 37.8 \/ 9.4 points appear in text\u2013image, image\u2013text and image\u2013image settings.<\/li><\/ul>\n\n\n<p class=\"wp-block-paragraph\">Compositional binding \u2014 &#8220;glowing giraffe&#8221; vs &#8220;glowing submarine&#8221; \u2014 still trips vision\u2013language models. OriginAI, NVIDIA and Tel Aviv University introduce SABER, which reassigns fixed object\/attribute sets to build hard distractors.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1344\" height=\"768\" src=\"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-acc.webp\" alt=\"Dumbbell: mean accuracy rises from 52.5 to 84.2 percent after text-only finetuning.\" class=\"wp-image-6167\" srcset=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-acc.webp 1344w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-acc-300x171.webp 300w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-acc-1024x585.webp 1024w, https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-acc-768x439.webp 768w\" sizes=\"auto, (max-width: 1344px) 100vw, 1344px\" \/><figcaption class=\"wp-element-caption\">i10X chart based on data from Tamir et al. (2026), arXiv:2610.22602, main finetuning results (CC BY-NC-SA 4.0).<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2c What the study shows<\/h2>\n\n\n<p class=\"wp-block-paragraph\">For most non-CLIP models, the correct candidate and three binding-swapped distractors occupy the top four retrieved slots in >90% of queries, so ranking reduces to binding. CLIP ViT-B\/32 variants only do so 27\u201353% of the time.<\/p>\n\n\n<ul class=\"wp-block-list\"><li><strong>2B model also improves:<\/strong> Average +15.6 points over its pretrained counterpart.<\/li><li><strong>Retrieval mAP:<\/strong> 8B model improves mean mAP by 19.2 points over Qwen3-VL-Embedding-8B.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u2705 Why it matters: composition is reliability<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Shopping, robotics and captioning fail when attributes attach to the wrong object. A text-only recipe that transfers cross-modally is cheap insurance.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83e\uddea Method in brief<\/h2>\n\n\n<p class=\"wp-block-paragraph\">SABER MCQ and retrieval protocols; LoRA adapters on frozen backbones; synthetic text supervision.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\u2753 What remains open<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Synthetic shuffles:<\/strong> Real-world binding errors may differ.<\/li><li><strong>Model coverage:<\/strong> Strongest gains on Qwen-line embeddings.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\ud83e\udded Context<\/h2>\n\n\n<p class=\"wp-block-paragraph\">Related VLM compositionality in <a href=\"\/news\/category\/studies\/multimodal\">Multimodal<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udee0\ufe0f What it means for embedding trainers<\/h2>\n\n<ul class=\"wp-block-list\"><li><strong>Add binding negatives:<\/strong> Shuffled attributes are high-value.<\/li><li><strong>Try text-only adaptation:<\/strong> Surprisingly transferable.<\/li><li><strong>Audit CLIP-style models:<\/strong> They miss the top-4 binding structure more often.<\/li><\/ul>\n\n\n<h2 class=\"wp-block-heading\">\u270d\ufe0f About the analysis<\/h2>\n\n\n<p class=\"wp-block-paragraph\">This article is based on the full text of the preprint &#8220;Stuck on Shuffle: On Attribute-Object Binding in Vision-Language Models&#8221; (arXiv:2610.22602v1, submitted October 15, 2026). All numbers are taken from the paper&#8217;s text, tables and figures; i10X did not rerun the experiments. The chart was redrawn by i10X from main finetuning results of the paper, which is published under a CC BY-NC-SA 4.0 license.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">\ud83d\udd2d i10x Perspective<\/h2>\n\n\n<p class=\"wp-block-paragraph\">That you can teach vision-language binding without showing a single image is both delightful and slightly worrying. It suggests the failure was linguistic structure in a shared space \u2014 and that we have been underusing text to fix multimodal bugs.<\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>SABER binding benchmark: an 8B model finetuned on synthetic text alone jumps from 52.5% to 84.2% mean accuracy, with +34.1 points text\u2013image \u2014 no visual training data.<\/p>\n","protected":false},"author":2,"featured_media":6318,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"paper_title":"Stuck on Shuffle: On Attribute-Object Binding in Vision-Language Models","paper_url":"https:\/\/arxiv.org\/abs\/2610.22602","pdf_url":"https:\/\/arxiv.org\/pdf\/2610.22602v1","doi":"","arxiv_id":"2610.22602","authors":"Netanel Y. Tamir, Michael Green, Issar Tzachor, Or Shimshi, Dvir Samuel, Or Patashnik, Rami Ben-Ari","institutions":"OriginAI; NVIDIA; Tel Aviv University","paper_date":"2026-10-15","venue":"arXiv (cs.CV), version 1","peer_reviewed":false,"code_url":"","data_url":"","funding_coi":"NVIDIA affiliation.","figure_license":"Chart drawn by i10X from main finetuning results (CC BY-NC-SA 4.0)","paper_license":"CC BY-NC-SA 4.0","replication":"no","study_models":"Qwen3-VL-Embedding 2B\/8B and CLIP variants with LoRA finetunes","study_sample":"SABER attribute\u2013object binding MCQ and retrieval","footnotes":"","_yoast_wpseo_primary_category":""},"categories":[58,52],"tags":[],"companies":[44],"products":[],"class_list":["post-6168","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-multimodal","category-studies","company-nvidia"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News<\/title>\n<meta name=\"description\" content=\"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News\" \/>\n<meta property=\"og:description\" content=\"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm\" \/>\n<meta property=\"og:site_name\" content=\"i10X News\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-19T16:49:43+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-10-19T17:24:20+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1344\" \/>\n\t<meta property=\"og:image:height\" content=\"768\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"Christopher Ort\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:site\" content=\"@i10x_ai\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Christopher Ort\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm\"},\"author\":{\"name\":\"Christopher Ort\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\"},\"headline\":\"VLMs Still Can&#8217;t Tell Which Attribute Goes With Which Object \u2014 Until Text-Only Finetuning\",\"datePublished\":\"2026-10-19T16:49:43+00:00\",\"dateModified\":\"2026-10-19T17:24:20+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm\"},\"wordCount\":387,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp\",\"articleSection\":[\"Multimodal\",\"Studies\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm\",\"name\":\"Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp\",\"datePublished\":\"2026-10-19T16:49:43+00:00\",\"dateModified\":\"2026-10-19T17:24:20+00:00\",\"description\":\"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp\",\"width\":1344,\"height\":768,\"caption\":\"Illustration: i10X\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/stuck-on-shuffle-attribute-object-binding-vlm#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/news.i10x.ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"VLMs Still Can&#8217;t Tell Which Attribute Goes With Which Object \u2014 Until Text-Only Finetuning\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#website\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"name\":\"i10X News\",\"description\":\"Stay updated with the latest news and announcements from i10X.\",\"publisher\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/news.i10x.ai\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#organization\",\"name\":\"i10X\",\"url\":\"https:\\\/\\\/news.i10x.ai\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"contentUrl\":\"https:\\\/\\\/i10xnews.kinsta.cloud\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/i10x-logo-180.png\",\"width\":180,\"height\":180,\"caption\":\"i10X\"},\"image\":{\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/x.com\\\/i10x_ai\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/i10x-ai\",\"https:\\\/\\\/instagram.com\\\/i10x.ai\",\"https:\\\/\\\/www.youtube.com\\\/@i10X_ai_official\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/news.i10x.ai\\\/#\\\/schema\\\/person\\\/d410bb7ee72febefca2c9c369939219a\",\"name\":\"Christopher Ort\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"contentUrl\":\"https:\\\/\\\/i10x.ai\\\/news\\\/wp-content\\\/uploads\\\/2026\\\/10\\\/christopher-ort-192.webp\",\"caption\":\"Christopher Ort\"},\"description\":\"Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.\",\"url\":\"https:\\\/\\\/i10x.ai\\\/news\\\/author\\\/christopher-ort\",\"sameAs\":[\"https:\\\/\\\/i10x.ai\\\/blog\\\/author\\\/christopher-ort\"],\"worksFor\":{\"@id\":\"https:\\\/\\\/i10x.ai\\\/news\\\/#organization\"},\"alumniOf\":{\"@type\":\"EducationalOrganization\",\"name\":\"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)\",\"url\":\"https:\\\/\\\/www.fau.eu\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News","description":"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm","og_locale":"en_US","og_type":"article","og_title":"Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News","og_description":"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.","og_url":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm","og_site_name":"i10X News","article_published_time":"2026-10-19T16:49:43+00:00","article_modified_time":"2026-10-19T17:24:20+00:00","og_image":[{"width":1344,"height":768,"url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp","type":"image\/webp"}],"author":"Christopher Ort","twitter_card":"summary_large_image","twitter_creator":"@i10x_ai","twitter_site":"@i10x_ai","twitter_misc":{"Written by":"Christopher Ort","Est. reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#article","isPartOf":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm"},"author":{"name":"Christopher Ort","@id":"https:\/\/news.i10x.ai\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a"},"headline":"VLMs Still Can&#8217;t Tell Which Attribute Goes With Which Object \u2014 Until Text-Only Finetuning","datePublished":"2026-10-19T16:49:43+00:00","dateModified":"2026-10-19T17:24:20+00:00","mainEntityOfPage":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm"},"wordCount":387,"commentCount":0,"publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"image":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp","articleSection":["Multimodal","Studies"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#respond"]}]},{"@type":"WebPage","@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm","url":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm","name":"Stuck on Shuffle: Fixing Attribute\u2013Object Binding in VLMs - i10X News","isPartOf":{"@id":"https:\/\/news.i10x.ai\/#website"},"primaryImageOfPage":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage"},"image":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage"},"thumbnailUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp","datePublished":"2026-10-19T16:49:43+00:00","dateModified":"2026-10-19T17:24:20+00:00","description":"SABER shows VLMs fail attribute\u2013object binding. Text-only LoRA finetuning lifts 8B mean accuracy from 52.5% to 84.2%.","breadcrumb":{"@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#primaryimage","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/stuck-on-shuffle-attribute-object-binding-vlm-title.webp","width":1344,"height":768,"caption":"Illustration: i10X"},{"@type":"BreadcrumbList","@id":"https:\/\/i10x.ai\/news\/stuck-on-shuffle-attribute-object-binding-vlm#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/news.i10x.ai\/"},{"@type":"ListItem","position":2,"name":"VLMs Still Can&#8217;t Tell Which Attribute Goes With Which Object \u2014 Until Text-Only Finetuning"}]},{"@type":"WebSite","@id":"https:\/\/news.i10x.ai\/#website","url":"https:\/\/news.i10x.ai\/","name":"i10X News","description":"Stay updated with the latest news and announcements from i10X.","publisher":{"@id":"https:\/\/news.i10x.ai\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/news.i10x.ai\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/news.i10x.ai\/#organization","name":"i10X","url":"https:\/\/news.i10x.ai\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/news.i10x.ai\/#\/schema\/logo\/image\/","url":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","contentUrl":"https:\/\/i10xnews.kinsta.cloud\/wp-content\/uploads\/2026\/10\/i10x-logo-180.png","width":180,"height":180,"caption":"i10X"},"image":{"@id":"https:\/\/news.i10x.ai\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/i10x_ai","https:\/\/www.linkedin.com\/company\/i10x-ai","https:\/\/instagram.com\/i10x.ai","https:\/\/www.youtube.com\/@i10X_ai_official"]},{"@type":"Person","@id":"https:\/\/news.i10x.ai\/#\/schema\/person\/d410bb7ee72febefca2c9c369939219a","name":"Christopher Ort","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","url":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","contentUrl":"https:\/\/i10x.ai\/news\/wp-content\/uploads\/2026\/10\/christopher-ort-192.webp","caption":"Christopher Ort"},"description":"Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU). He is the author of \u201cThe Style Penalty\u201d, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.","url":"https:\/\/i10x.ai\/news\/author\/christopher-ort","sameAs":["https:\/\/i10x.ai\/blog\/author\/christopher-ort"],"worksFor":{"@id":"https:\/\/i10x.ai\/news\/#organization"},"alumniOf":{"@type":"EducationalOrganization","name":"Friedrich-Alexander-Universit\u00e4t Erlangen-N\u00fcrnberg (FAU)","url":"https:\/\/www.fau.eu"}}]}},"_links":{"self":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6168","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/comments?post=6168"}],"version-history":[{"count":2,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6168\/revisions"}],"predecessor-version":[{"id":6319,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/posts\/6168\/revisions\/6319"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media\/6318"}],"wp:attachment":[{"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/media?parent=6168"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/categories?post=6168"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/tags?post=6168"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/companies?post=6168"},{"taxonomy":"product","embeddable":true,"href":"https:\/\/i10x.ai\/news\/wp-json\/wp\/v2\/products?post=6168"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}