arXiv:2609.39168v1 Announce Type: new
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has improved the reasoning capabilities of Multimodal Large Language Models (MLLMs), yet existing...
By Zhihan Zhang, Lizi Liao
The paper introduces VIG (Visual Information Gain), an information‑theoretic reward that evaluates each token in a multimodal chain‑of‑thought by measuring how much the image reduces its predictive uncertainty. VIG is computed online using two forward passes—one with and one without the image—eliminating the need for reference chains or external annotations. Experiments on six multimodal reasoning benchmarks and multiple Qwen3‑VL‑Thinking model sizes show that VIG consistently improves the accuracy–efficiency trade‑off, demonstrating that efficient multimodal reasoning arises from increasing visual information density rather than merely limiting chain length.
By Wen Luo, Xiaohan Yi, Xiaotao Huang, Liqun Huang
The paper introduces RECAP, a training method that enhances vision‑language models by incorporating temporally localized reactivation of image evidence—called visual revisit—into the reinforcement learning process. Visual revisit peaks are shown to predict correctness and causally improve performance; masking them hurts accuracy more than masking non‑peaks, while adding revisit peaks to failed reasoning partially restores correct answers. RECAP applies revisit‑conditioned advantage estimation, dynamic discounting, and gated value prediction to propagate final rewards through visually important reasoning steps, achieving significant gains on HallusionBench and other benchmarks without extra rewards or inference‑time changes.
By Zhaolu Kang, Tailong Luo, Chenxin Li, Yingjie He, Zhenyu Yu, Fengyu Zhou, Jiachen Qian, Lei Wei, Shuang Chen, Jiachen Li, Eric Hanchen Jiang, Rongchao Zhang, Zhengtao Yao, Hoi Leong Lee, Guansu Wang, Kaiyue Zhou
arXiv:2604.13993v2 Announce Type: replace-cross
Abstract: Physical reasoning over images requires integrating perception, domain knowledge, and multi-step symbolic inference, yet Vision-Language Mode...
By Derek Lilienthal, Manisha Mukherjee, Sameera Horawalavithana
arXiv:2605. 14054v2 Announce Type: replace Abstract: Achieving robust perception-reasoning synergy is a central goal for advanced Vision-Language Models (VLMs).
By Haozhe Wang, Qixin Xu, Changpeng Wang, Taofeng Xue, Chong Peng, Wenhu Chen, Fangzhen Lin
arXiv:2603. 06652v2 Announce Type: replace-cross Abstract: Reinforcement learning has recently improved the reasoning ability of Large Language Models and Multimodal LLMs, yet prevailing reward designs emphasise final-answer correctness and consequently tolerate process hallucinations--cases where models reach the right answer while misperceiving visual evidence.
By Yantao Li, Qiang Hui, Chenyang Yan, Kanzhi Cheng, Fang Zhao, Chao Tan, Huanling Gao, Jianbing Zhang, Kai Wang, Xinyu Dai, Shiguo Lian
The paper demonstrates that reinforcement learning with verifiable rewards (RLVR) can improve vision‑language benchmark performance even when models are trained without visual input. When real images are introduced at test time, models trained blind recover about half of the performance gain at 3B parameters and nearly four‑fifths at 7B, but extended real‑image training can erode grounding while benchmark gains persist. The authors propose a visual resolvability rule and show that requiring visual evidence for correct answers leads to significant improvements in target discovery and generalization to unseen question types, while controls confirm that the gains stem from actual visual grounding rather than artifacts.
By Haocun Ye, Xinlong Jiang, Qile Chen, Bingyu Wang, Teng Zhang, Shubai Chen, Tingyu Wu, Zhenkun Zheng, Yiqiang Chen
PerceptRL enhances visual perception in multimodal large language models by employing reinforcement learning that transforms dense image captions into verified QA pairs with rule‑based rewards. It uses a multi‑agent pipeline to link correlated errors, combine unrelated errors, and target high‑error domains, while a specialized multi‑turn inspection process corrects spatial inaccuracies. Across seven benchmarks, PerceptRL improves average accuracy to 71.6%, outperforming both the base model with tools and the leading evaluated VLM by significant margins.
By Tongkun Guan, Zhibo Yang, Mingkun Yang, Haocheng Wang, Wei Shen, Xiaokang Yang
arXiv:2602.08241v2 Announce Type: replace
Abstract: While chain-of-thought (CoT) reasoning has substantially improved multimodal large language models (MLLMs) on complex reasoning tasks, existing app...
By Siqu Ou, Tianrui Wan, Zhiyuan Zhao, Junyu Gao, Xuelong Li
The paper examines how current reward models used in text-to-image systems struggle with visual reasoning tasks. By introducing a data‑centric approach that flips the pairing axis—using visual reasoning datasets as preference pairs with the same image but different prompts—the authors achieve simultaneous gains in visual understanding and preference accuracy. This method yields measurable improvements in generation quality on compositional prompts, outperforming a larger baseline model.
By Shiye Su, Xiaohan Wang, Ludwig Schmidt, Serena Yeung-Levy
UniCAR‑RL is an annotation‑free reinforcement learning framework designed to improve multimodal large language models’ visual mathematics reasoning. It decouples perception and reasoning by using three branches: Caption‑RL for perception optimization, Reasoning‑RL for logical reasoning with a gold image description, and QA‑RL for end‑to‑end question answering. Experiments show significant gains in mathematical and visual reasoning across various model architectures and scales using only raw short‑answer data.
By Yuzhe Li, Hao Yan, Hao Wang, Xingchen Liu, Ya-Qi Yu, Jihao Wu, Minghui Liao, Wei Chen, Yuliang Liu
The paper investigates the faithfulness of visual information in multimodal chain-of-thought (MCoT) reasoning by large vision-language models (LVLMs). It shows that visual cues in MCoT are often inaccurate and largely ignored, as predictions change little when visual input is altered but significantly when textual input is changed. To improve visual faithfulness, the authors propose an annotation‑free Sufficient‑Component Cause Model (SCCM) that encourages the generation of minimal yet sufficient visual components, and demonstrate its effectiveness across fine‑grained perception and reasoning benchmarks.
By Zujing Liu, Junwen Pan, Qi She, Yuan Gao, Guisong Xia