[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86093-en":3,"doc-seo-86093-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},86093,2336464648746,"Skyler","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","SVR-R1 Bootstrapping Multi-modal Reasoning with Self-verification in Reinforcement Learning","SVR-R1 presents Self-Verified Reasoner as a multi-turn reinforcement learning framework for multimodal reasoning that uses a model’s own verification as a training signal. For each query, the model generates an answer with shared weights, then emits a binary self-verdict (Yes/No). A No verdict triggers a second-chance rethink; a Yes (or turn limit) finalizes output for outcome-based reward. Implemented with GRPO and asynchronous multi-turn rollouts, SVR-R1 requires no external supervision or auxiliary critics, and improves vision-language reasoning accuracy substantially over strong GRPO baselines.","arXiv :2607 . 10966v 1 [ cs .AI] 13 Jul 2026  \nSVR-R1: Bootstrapping Multi-modal Reasoning with Selfverification in Reinforcement Learning  \nMingyuan Wu1, Jingcheng Yang1, Shengyi Qian2, Xudong Wang2, Jize Jiang1, Qifan Wang2, Aashu Singh2, Khoi Pham2, Fei Liu2, Zhaolun Su2,  \nZhuokai Zhao2, Klara Nahrstedt1, Jianyu Wang2, Hanchao Yu2  \n1University of Illinois Urbana-Champaign, 2Meta {[mw34](mw34}@illinois.edu)[}](mw34}@illinois.edu)[@illinois.edu](mw34}@illinois.edu)  \nAbstract  \nWe introduce Self-Verified Reasoner (SVR-R1), a multi-turn RL framework that turns a model’s own verification into a learning signal for multimodal reasoning. For each query, the model proposes an answer using the same weights, and issues a binary self-verdict (Yes/No) . A “No” triggers a second-chance rethink; a “Yes,” or a turn cap, finalizes the output for computing the outcome-based reward. SVR-R1 is implemented with GRPO and an asynchronous multi-turn rollout framework and needs no external supervision or auxiliary critics. We evaluate SVR-R1 on visionlanguage reasoning benchmarks and show that it improves accuracy by a large margin over strong standard GRPO baselines. Training dynamics show decreasing reliance on verification-fewer verification turns, yet higher test accuracy-indicating that the gap between verification and generation narrows as the policy internalizes self-correction and chooses the most confident answer via our framework. SVR-R1 bridges the less explored intersection of inference-time self-refinement and RL training for VLMs, offering a simple yet effective recipe for bootstrapping multimodal reasoning.  \nWe will open-source SVR-R1 to facilitate future research in VLMs.  \n1 Introduction  \nWhen given a second chance to think, humans often reason their way toward better solutions on complex problems. Recently, Large Language Models (LLMs) have demonstrated a similar pattern of iterative rethinking in the reasoning process (OpenAI, 2024a), particularly when fine-tuned with reinforcement learning (RL) on task-specific rewards (DeepSeek  \nAI, 2025; Zeng et al., 2025; Zhou et al.,  \n2025b; Wang et al., 2025b) . After such training, these models can intrinsically exhibit reasoning behaviors that allow them to iteratively improve their reasoning paths through behaviors like self-correction and backtracking (Gandhi et al., 2025) . This approach was initially applied to verifiable tasks in the language domain, such as mathematics and coding, and has since been extended to non-verifiable domains and to vision-language tasks (Zhou et al., 2025a; Chen et al., 2025a; Zhang et al., 2025; Huang et al., 2025a; Deng et al., 2025; Wang et al., 2025a) .  \nEven before the widespread adoption of RL fine-tuning for enabling self-rethinking in LLMs (Kumar et al., 2025), it was observed that simply prompting an LLM to second-guess to refine or correct its previous answer could lead to improved reasoning performance in inference (Madaan et al., 2023; Shinn et al., 2023) . This improvement may be attributed to the gap between verification and generation (Song et al., 2025): “if verification is easier  \nFigure 1: Merging VLM Self-Verification Loop into RL Rollout  \nthan generation, one can hypothesize that the model may act as a better-than-random verifier of its own outputs, enabling self-improvement”. Beyond pure language, pioneering work (Liao et al., 2025) has shown that strong commercial Vision Language Models (VLMs) such as GPT-4o can, to some extent, self-verify their outputs in certain tasks (e.g. segmentation), though effective self-rethinking generally remains challenging and underexplored in multi-modal reasoning scenarios Wu et al. (2025); Huang et al. (2025b) .  \nIn this work, we explore a previously unexplored middle ground between explicit prompting and RL fine-tuning for self-rethinking by leveraging VLMs’ pre-existing self-verification capabilities in Figure 1. Specifically, we aim to bootstrap a VLM’s reasoning abilities, prompting it ","cbCaivtFhsnlDiYt","https://ap.wps.com/l/cbCaivtFhsnlDiYt","pdf",2263916,2,1,18,"English","en",105,"# Abstract\n# Introduction\n# Method Overview\n# Experiments and Results\n# Training Dynamics and Analysis\n# Open-Source Plan","[{\"question\":\"How does SVR-R1 use self-verification during reinforcement learning rollouts?\",\"answer\":\"For each query, the model generates a tentative answer and then issues a binary self-verdict (Yes/No). A No verdict causes the model to rethink and regenerate, while a Yes (or reaching a turn cap) finalizes the output for reward computation.\"},{\"question\":\"What training components does SVR-R1 rely on?\",\"answer\":\"SVR-R1 is trained using GRPO with outcome-based rewards and an asynchronous multi-modal multi-turn rollout framework to coordinate the generator and self-verifier.\"},{\"question\":\"What improvements does SVR-R1 achieve on vision-language reasoning benchmarks?\",\"answer\":\"Evaluation on vision-language reasoning benchmarks shows accuracy gains by a large margin over strong GRPO baselines, indicating that self-verification integrated into RL training improves multimodal reasoning performance.\"}]",1784208459,45,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"svr-r1-bootstrapping-multi-modal-reasoning-with-self-verification-in-reinforcement-learning","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/svr-r1-bootstrapping-multi-modal-reasoning-with-self-verification-in-reinforcement-learning/86093/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-26","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"How does SVR-R1 use self-verification during reinforcement learning rollouts?","Question",{"text":75,"@type":76},"For each query, the model generates a tentative answer and then issues a binary self-verdict (Yes/No). A No verdict causes the model to rethink and regenerate, while a Yes (or reaching a turn cap) finalizes the output for reward computation.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What training components does SVR-R1 rely on?",{"text":80,"@type":76},"SVR-R1 is trained using GRPO with outcome-based rewards and an asynchronous multi-modal multi-turn rollout framework to coordinate the generator and self-verifier.",{"name":82,"@type":73,"acceptedAnswer":83},"What improvements does SVR-R1 achieve on vision-language reasoning benchmarks?",{"text":84,"@type":76},"Evaluation on vision-language reasoning benchmarks shows accuracy gains by a large margin over strong GRPO baselines, indicating that self-verification integrated into RL training improves multimodal reasoning performance.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]