Event understanding and reasoning play critical roles in thoroughly evaluating the capabilities of Vision-Language Models (VLMs); however, existing Visual Question Answering (VQA) datasets predominantly focus on entity-centric questions, while event- or action-related questions are limited in scale and suffer from significant shortcut issues. We introduce MEUR, the first Multimodal Event Understanding and Reasoning dataset consisting of 1,200 images and 4,217 questions, necessitating VLMs with a diverse range of multimodal understanding and reasoning capabilities to answer, ranging from basic event recognition to more complex tasks such as counting and comparison. To streamline the annotation process, we propose a novel semi-automated pipeline that combines advanced VLMs with human annotators, achieving high quality and efficiency. We conduct extensive experiments on state-of-the-art non-thinking and thinking VLMs to demonstrate their capabilities and limitations in multimodal event understanding and reasoning. Furthermore, we provide a detailed error analysis that points out promising directions for future research.

To build a math MLLM, current research efforts predominantly rely on elaborate training schemes or high-quality multi-modal math reasoning data. On the other hand, model merging has attracted much attention recently due to its low-cost and tune-free property, which has shown great potential for constructing math MLLM. However, most recent attempts either assume all parameters are task-relevant or adopt a less interpretable parameter selection strategy, thus merging parameters on a coarse granularity, leading to limited or even degrading performance. In this paper, we propose a causal-analysis–guided parameter merging framework, which identifies the math reasoning-related parameters in a finer granularity and interpretable manner, then merges the math LLMs with MLLM. Specifically, we first construct targeted arithmetic and logical tasks, generate reference and counterfactual samples, and then apply intervention-based causal attribution to localize parameters that are causally responsible for math reasoning. Once we identify the crucial reasoning parameters in the math LLM that exhibit higher causal effects, we can more precisely align and merge only the corresponding parameters from the math LLM into the base MLLM. The procedure is tuning-free and enhances both math reasoning and general reasoning abilities. Experimental results of three model families on four math reasoning benchmarks and two general reasoning benchmarks validate the effectiveness of the proposed approach.

Cultural image retrieval is essential for organizing and analyzing large-scale digital heritage resources. However, existing multimodal retrieval methods usually encode a query as a holistic embedding, making it difficult to preserve logical operators and cultural relations during retrieval. Therefore, we introduce Multimodal Boolean Logic Culture Retrieval, a novel task that moves beyond general visual similarity by enforcing strict Boolean constraints over structured cultural elements. To tackle this, we propose *CLAIR*, a training-free framework that reformulates the retrieval objective to seamlessly integrate logical operations directly within the multimodal embedding space. Furthermore, we construct CulTi2.0, a dedicated benchmark designed to evaluate diverse logical query forms based on cultural visual atoms. Extensive experiments show that while current multimodal models struggle with explicit logical queries, *CLAIR* provides consistent and generalizable improvements as a plug-and-play solution across various datasets and model families. Code and dataset are publicly available at https://github.com/AI4Culture/CulTi2.0.

The mid-training phase has emerged as a critical bridge for enhancing the reasoning capabilities of Vision-Language Models, yet its effectiveness is constrained by the scarcity of high-quality visual reasoning data. While data synthesis offers a promising alternative, conventional stateless synthesis paradigms produce shallow trajectories that quickly succumb to utility plateaus. To overcome this, we propose VisMind, a novel multi-agent collaborative framework that generates deep-reasoning Stateful Synthetic Dialogue (SSD) from visual inputs. By orchestrating iterative problem-solving, where peer student agents engage in bidirectional critique under the ground-truth-anchored guidance of a teacher, VisMind transforms static, fragmented responses into rich, multi-turn logical derivations. Utilizing this interactive paradigm, we construct a dense mid-training corpus of approximately 1.14B tokens. Extensive experiments across the Qwen2-VL and InternVL3 model families demonstrate that applying this corpus via Continued Pretraining consistently outperforms a representative stateless synthesis approach and elevates standard baselines by up to a 2.24% absolute average gain across comprehensive mathematical and logical benchmarks. Furthermore, in-depth analysis reveals that the SSD paradigm exhibits continuous linear scaling without early saturation, fosters intrinsic reasoning generalization, and establishes a robust cognitive foundation that enhances the efficacy of subsequent Reinforcement Learning with Verifiable Rewards.

Dunhuang art, a cornerstone of global heritage, demands fine-grained visual perception anchored by specialized cultural knowledge. Given the strong performance of multimodal large language models (MLLMs) on generic multimodal benchmarks, to what extent can they understand artifacts from Dunhuang art that are grounded in cultural context? To this end, we construct Dunhuang-Bench, a large-scale benchmark comprising 486 images and 22,970 QA pairs. It incorporates diverse task formats to evaluate MLLMs’ cultural understanding: Question Answering with Text Description, Multi-turn Dialogue, and Question Answering with Choices. Guided by Panofsky’s theory of iconology, we design two tasks including visual perception and knowledge reasoning for the evaluation of content understanding. In addition, we follow the theory of formal analytic tradition to design another task of artistic appreciation in our Dunhuang-Bench. Extensive evaluations of 20 mainstream MLLMs on Dunhuang-Bench reveal a consistent performance drop from perception and appreciation to reasoning. Moreover, CoT and few-shot prompting show marginal or negative impact, highlighting the limits of prompting-based improvements. Dunhuang-Bench thus provides a challenging benchmark for advancing multimodal cultural understanding. Data and code will be publicly available.

Plane geometry problem (PGP) solving has become a critical benchmark for multimodal reasoning because it requires accurate visual perception and precise multi-step symbolic deduction. Although test-time scaling (TTS) has demonstrated remarkable success in general mathematical reasoning, it fails to scale effectively under the symbolic-program paradigm for plane geometry. We identify two key obstacles: limited reasoning diversity induced by rigid symbolic programs and insufficient explicit visual grounding before symbolic deduction. To address these issues, we propose Multi-Trace Synthesis (MTS), which converts each symbolic program into heterogeneous reasoning traces, including executable Python scripts and CoT-augmented variants. We further propose Perception-Augmented (PA) training, which parses diagrams into structured semantic clauses before deduction, and Consensus-Guided Multi-Trace Ensemble (CG-MTE) for efficient self-adaptive inference. Experiments on three geometry benchmarks show that our method consistently improves PGP-solving across model scales and achieves strong performance against both general-purpose MLLMs and specialized geometry solvers. Under test-time scaling, CG-MTE achieves a comparable accuracy to high-budget self-consistency while reducing sampling cost by up to 8×. Code and data are publicly available at https://github.com/Jason8Kang/ReTTS-PGPS.

Cross-modal retrieval is essential for interpreting cultural heritage data, but its effectiveness is often limited by incomplete or inconsistent textual descriptions, caused by historical data loss and the high cost of expert annotation. While large language models (LLMs) offer a promising solution by enriching textual descriptions, their outputs frequently suffer from hallucinations or miss visually grounded details. To address these challenges, we propose C^3, a data augmentation framework that enhances cross-modal retrieval performance by improving the completeness and consistency of LLM-generated descriptions. C^3 introduces a bidirectional validation mechanism to assess semantic coverage using both visual cues and language-model outputs. Furthermore, to mitigate factual inconsistencies, we formulate a Markov Decision Process to supervise Chain-of-Thought reasoning, guiding consistency verification through adaptive query control. Experiments on the cultural heritage dataset CulTi and general benchmarks MSCOCO and Flickr30K demonstrate that C^3 achieves state-of-the-art performance in both fine-tuned and zero-shot settings.

Existing table understanding methods face challenges due to complex table structures and intricate logical reasoning. While supervised finetuning (SFT) dominates existing research, reinforcement learning (RL), such as Group Relative Policy Optimization (GRPO), has shown promise but struggled with low initial policy accuracy and coarse rewards in tabular contexts. In this paper, we introduce Table-R1, a three-stage RL framework that enhances multimodal table understanding through: (1) Warm-up that prompts initial perception and reasoning capabilities, (2) Perception Alignment GRPO (PA-GRPO), which employs continuous Tree-Edit-Distance Similarity (TEDS) rewards for recognizing table structures and contents, and (3) Hint-Completion GRPO (HC-GRPO), which utilizes fine-grained rewards of residual steps based on the hint-guided question. Extensive experiments demonstrate that Table-R1 can boost the model’s table reasoning performance obviously on both held-in and held-out datasets, outperforming SFT and GRPO largely. Notably, Qwen2-VL-7B with Table-R1 surpasses larger specific table understanding models (e.g., Table-LLaVA 13B), even achieving comparable performance to the closed-source model GPT-4o on held-in datasets, demonstrating the efficacy of each stage of Table-R1 in overcoming initialization bottlenecks and reward sparsity, thereby advancing robust multimodal table understanding.

Solving math word problem (MWP) with AI techniques has recently made great progress with the success of deep neural networks (DNN), but it is far from being solved. We argue that the ability of learning by analogy is essential for an MWP solver to better understand same problems which may typically be formulated in diverse ways. However most existing works exploit the shortcut learning to train MWP solvers simply based on samples with a single question. In lack of diverse questions, these methods merely learn shallow heuristics. In this paper, we make a first attempt to solve MWPs by generating diverse yet consistent questions/equations. Given a typical MWP including the scenario description, question, and equation (i.e., answer), we first generate multiple consistent equations via a group of heuristic rules. We then feed them to a question generator together with the scenario to obtain the corresponding diverse questions, forming a new MWP with a variety of questions and equations. Finally we engage a data filter to remove those unreasonable MWPs, keeping the high-quality augmented ones. To evaluate the ability of learning by analogy for an MWP solver, we generate a new MWP dataset (called DiverseMath23K) with diverse questions by extending the current benchmark Math23K. Extensive experimental results demonstrate that our proposed method can generate high-quality diverse questions with corresponding equations, further leading to performance improvement on Diverse-Math23K. The code and dataset is available at: https://github.com/zhouzihao501/DiverseMWP.