[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-273881-105":59,"doc-detail-273881-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","comet-contrastive-motion-enhanced-temporal-reasoning-for-video-multimodal-large-language-models","COMET - Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models","","Video multimodal large language models show strong performance, but fine-grained motion and temporal understanding remains fragile. The bottleneck goes beyond sparse frame sampling: many pipelines lack an end-to-end temporal modeling mechanism that explicitly represents frame-to-frame change, enables appearance–motion interaction, and captures temporal direction sensitivity. COMET introduces a temporally grounded framework with an explicit temporal motion branch and direction-aware optimization, improving action-centric and temporal reasoning benchmarks while keeping static perception on par.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/comet-contrastive-motion-enhanced-temporal-reasoning-for-video-multimodal-large-language-models/273881/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/comet-contrastive-motion-enhanced-temporal-reasoning-for-video-multimodal-large-language-models/273881.png","ImageObject",300,407,{"name":92,"@type":93},"Aurelia","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-02","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":56},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does COMET address in video multimodal large language models?","Question",{"text":112,"@type":113},"COMET targets fragile fine-grained motion-temporal understanding, caused by missing explicit frame-to-frame temporal representation and insufficient optimization for direction- or order-sensitive behavior.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does COMET represent temporal motion explicitly?",{"text":117,"@type":113},"COMET adds a temporal motion branch using Taylor-style frame differences, then injects motion evidence into the appearance stream through temporal attention bias-enhanced cross-attention.",{"name":119,"@type":110,"acceptedAnswer":120},"What optimization strategy does COMET use to improve temporal reasoning?",{"text":121,"@type":113},"COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage, turning temporal order into a direct learning signal and strengthening reliance on direction-sensitive motion patterns.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},273881,1789473129,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":56,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":44,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":129,"read_time":143,"preview_page_count":4},1099514068365,"https://ap-avatar.wpscdn.com/avatar/10000253d8d9f28188e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1794309523707217728","COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models  \narXiv : 2608 .2 1030v 1 [cs .CV] 2 1 Aug 2026  \nChenghua Zhu∗ Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology, Shenzhen Graduate School Peking University Shenzhen, China  \nKehan Jiang  \nPeking University Beijing, China  \nZhaolu Kang∗ School of Software and Microelectronics Peking University Beijing, China  \nLei Wei  \nPeking University Beijing, China  \nQifan Shi∗ School of Future Technology  \nSouth China University of Technology Guangzhou, China  \nLianyu Hu Nanyang Technological  \nUniversity Singapore, Singapore  \nSiyan Wu  \nAberdeen Institute of Data Science and Artificial Intelligence South China Normal University Foshan, China  \nGuangyuan Dong  \nNational University of Singapore Singapore, Singapore  \nMingbo Yang  \nSun Yat-Sen University Guangzhou, China  \nRui Lu  \nPingan Technology Shenzhen, China  \nGuibo Luo✉ Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology, Shenzhen Graduate School Peking University Shenzhen, China  \nAbstract  \nVideo multimodal large language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile. The core bottleneck is not only sparse frame sampling, but also the lack of a complete temporal modeling pipeline for explicitly representing frame-to-frame change, enabling appearance-motion interaction, and optimizing temporal direction sensitivity. We propose COMET, a temporally grounded framework that systematically strengthens video MLLMs through explicit temporal representation, appearance-motion fusion, and direction-aware optimization. Architecturally, COMET introduces a temporal motion branch built on Taylor frame differences and injects its motion evidence into the appearance stream via temporal attention bias-enhanced crossattention. For optimization, COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal and strengthens the model’suse of directional motion patterns encoded by the temporal motion branch. The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, actioncentric tasks (STAR, SSv2) improve by 4.9% on average, temporal reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par. The same gain pattern also transfers to InternVL2.5-8B, indicating that COMET generalizes across model families.  \n∗ These authors contributed equally to this work.  \nMM’26, Rio de Janeiro, Brazil 2026.  \nCCS Concepts  \n• Computing methodologies → Computer vision; Natural language processing; Reinforcement learning.  \nKeywords  \nmultimodal large language models, video reasoning, temporal understanding, video question answering  \n1 Introduction  \nVideo multimodal large language models (video MLLMs), represented by systems such as Qwen3-VL [1], InternVL2.5 [3], and VideoLLaMA 2 [4], have recently shown strong progress on video question answering, dense video understanding, and long-form reasoning. Yet their temporal reasoning remains fragile. Many systems answer correctly when static appearance cues or language priors are sufficient, but they still struggle when the answer depends on how a scene evolves over time, such as motion direction, action order, or changes in state. In such cases, temporally grounded evidence is often weaker than the final answer confidence.  \nWe argue that a major reason is that current video MLLMs still underuse video as a temporal signal. In many pipelines, videos are reduced to sparsely sampled static frames [17] that are then encoded independently, often as sparse frame tokens [2], before being compressed, pooled, or aligned with the language model. Explicit inter-frame interaction is absent, so temporal evidence becomes sparse and discontinuous. As a result, video data util","cbCaidOxg01qIYeU","https://ap.wps.com/l/cbCaidOxg01qIYeU","pdf",14046478,"English","# Introduction\n## Problem: fragile temporal reasoning\n## Approach: temporally grounded COMET\n# Method Overview\n## Temporal motion branch with Taylor frame differences\n## Appearance–motion fusion via temporal attention bias\n## Optimization with temporal contrastive GRPO\n# Experiments and Results","[{\"question\":\"What problem does COMET address in video multimodal large language models?\",\"answer\":\"COMET targets fragile fine-grained motion-temporal understanding, caused by missing explicit frame-to-frame temporal representation and insufficient optimization for direction- or order-sensitive behavior.\"},{\"question\":\"How does COMET represent temporal motion explicitly?\",\"answer\":\"COMET adds a temporal motion branch using Taylor-style frame differences, then injects motion evidence into the appearance stream through temporal attention bias-enhanced cross-attention.\"},{\"question\":\"What optimization strategy does COMET use to improve temporal reasoning?\",\"answer\":\"COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage, turning temporal order into a direct learning signal and strengthening reliance on direction-sensitive motion patterns.\"}]","COMET - Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models | PDF",23]