Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a leading paradigm for enhancing visual reasoning in Multimodal Large Language Models (MLLMs). However, existing RLVR methods optimize primarily for the reasoning outcome and overlook how that outcome is produced: as a chain-of-thought unfolds, some tokens must extract evidence from the image while others must anchor to the preceding textual context, yet a single sequence-level reward provides no signal about whether each token draws on the modality its role demands. Token-level analyses and controlled interventions reveal that tokens frequently under-attend to their designated modality, and that this coordination breakdown may contribute to reasoning failures. Motivated by these findings, we propose DyCo-RL, which integrates dynamic cross-modal coordination into RLVR optimization. Specifically, DyCo-RL uses the Fisher-Rao geodesic distance to measure within-modality attention shifts, assigning tokens to either visually-oriented or text-oriented functional roles. It then evaluates the alignment between a token's actual attention allocation and its assigned role, leveraging this score for alignment-guided advantage reweighting during policy optimization. Extensive experiments using 11K examples for RL training demonstrate that the algorithm-agnostic DyCo-RL, when applied to Qwen2.5-VL-3B/7B and Qwen3.5-27B, improves the average performance of all four evaluated RLVR algorithms across seven benchmarks spanning visual-centric and mathematical reasoning.
VLA-Mark: A cross modal watermark for large vision-language alignment model
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；The Hong Kong University of Science and Technology（香港科技大学）；University of Toronto（多伦多大学）；Ant Group, Alibaba（蚂蚁集团，阿里巴巴）；New York University Shanghai（纽约大学上海分校）
Gwangju Institute of Science and Technology（光州科学技术院）
Imitation learning is employed to learn sensorimotor coordination for steering angle prediction in an end-to-end fashion requires expert demonstrations. These expert demonstrations are paired with environmental perception and vehicle control data. The conventional frame-based RGB camera is the most common exteroceptive sensor modality used to acquire the environmental perception data. The frame-based RGB camera has produced promising results when used as a single modality in learning end-to-end lateral control. However, the conventional frame-based RGB camera has limited operability in illumin…
ThinkMorph: Emergent Properties in Multimodal Interleaved Chain-of-Thought Reasoning
National University of Singapore（新加坡国立大学）；Zhejiang University（浙江大学）；University of Washington（华盛顿大学）；Stanford University（斯坦福大学）；absolute AI；The Chinese University of Hong Kong（香港中文大学）
ORCA: Orchestrated Reasoning with Collaborative Agents for Document Visual Question Answering
Digital Research Center of Sfax, SMARTS Laboratory（突尼斯斯法克斯数字研究中心，SMARTS实验室）；École Polytechnique de Tunisie, University of Carthage（突尼斯理工学院，卡塔赫纳大学）；Computer Vision Center, Universitat Autònoma de Barcelona（巴塞罗那自治大学计算机视觉中心）
MMoT: Mixture-of-Modality-Tokens Transformer for Composed Multimodal Conditional Image Synthesis
South China University of Technology（华南理工大学）；JD Explore Academy（京东探索研究院）；Nanyang Technological University（南洋理工大学）；Shanghai Jiao Tong University（上海交通大学）
Walking the Tightrope: Disentangling Beneficial and Detrimental Drifts in Non-Stationary Custom-Tuning
Australian Artificial Intelligence Institute (AAII)（澳大利亚人工智能研究所）；Faulty of Engineering and Information Technology（工程与信息技术学院）；University of Technology Sydney（悉尼技术大学）
Lappeenranta-Lahti University of Technology LUT（拉佩宁拉-拉赫蒂技术大学）；Finnish Environment Institute（芬兰环境研究所）；Faculty of Information Technology（信息科技学院）；Brno University of Technology（布拉格技术大学）
VividListener: Expressive and Controllable Listener Dynamics Modeling for Multi-Modal Responsive Interaction
Beijing University of Posts and Telecommunications（北京邮电大学）；Hong Kong University of Science and Technology（香港科技大学）；Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）
