Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning

Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.

MT$^{3}$: Scaling MLLM-based Text Image Machine Translation via Multi-Task Reinforcement Learning

Text Image Machine Translation (TIMT)-the task of translating textual content embedded in images-is critical for applications in accessibility, cross-lingual information access, and real-world document understanding. However, TIMT remains a complex challenge due to the need for accurate optical character recognition (OCR), robust visual-text reasoning, and high-quality translation, often requiring cascading multi-stage pipelines. Recent advances in large-scale Reinforcement Learning (RL) have improved reasoning in Large Language Models (LLMs) and Multimodal LLMs (MLLMs), but their application to end-to-end TIMT is still underexplored. To bridge this gap, we introduce MT$^{3}$, the first framework to apply Multi-Task RL to MLLMs for end-to-end TIMT. MT$^{3}$ adopts a multi-task optimization paradigm targeting three key sub-skills: text recognition, context-aware reasoning, and translation. It is trained using a novel multi-mixed reward mechanism that adapts rule-based RL strategies to TIMT's intricacies, offering fine-grained, non-binary feedback across tasks. Furthermore, to facilitate the evaluation of TIMT in authentic cross-cultural and real-world social media contexts, we introduced XHSPost, the first social media TIMT benchmark. Our MT$^{3}$-7B-Zero achieves state-of-the-art results on the latest in-domain MIT-10M benchmark, outperforming strong baselines such as Qwen2.5-VL-72B and InternVL2.5-78B by notable margins across multiple metrics. Additionally, the model shows strong generalization to out-of-distribution language pairs and datasets. In-depth analyses reveal how multi-task synergy, reinforcement learning initialization, curriculum design, and reward formulation contribute to advancing MLLM-driven TIMT.

CompBench: Benchmarking Complex Instruction-guided Image Editing

While real-world applications increasingly demand intricate scene manipulation, existing instruction-guided image editing benchmarks often oversimplify task complexity and lack comprehensive, fine-grained instructions. To bridge this gap, we introduce, a large-scale benchmark specifically designed for complex instruction-guided image editing. CompBench features challenging editing scenarios that incorporate fine-grained instruction following, spatial and contextual reasoning, thereby enabling comprehensive evaluation of image editing models' precise manipulation capabilities. To construct CompBench, We propose an MLLM-human collaborative framework with tailored task pipelines. Furthermore, we propose an instruction decoupling strategy that disentangles editing intents into four key dimensions: location, appearance, dynamics, and objects, ensuring closer alignment between instructions and complex editing requirements. Extensive evaluations reveal that CompBench exposes fundamental limitations of current image editing models and provides critical insights for the development of next-generation instruction-guided image editing systems.

Large Language Model Enhancers for Graph Neural Networks: An Analysis from the Perspective of Causal Mechanism Identification

The use of large language models (LLMs) as feature enhancers to optimize node representations, which are then used as inputs for graph neural networks (GNNs), has shown significant potential in graph representation learning. However, the fundamental properties of this approach remain underexplored. To address this issue, we propose conducting a more in-depth analysis of this issue based on the interchange intervention method. First, we construct a synthetic graph dataset with controllable causal relationships, enabling precise manipulation of semantic relationships and causal modeling to provide data for analysis. Using this dataset, we conduct interchange interventions to examine the deeper properties of LLM enhancers and GNNs, uncovering their underlying logic and internal mechanisms. Building on the analytical results, we design a plug-and-play optimization module to improve the information transfer between LLM enhancers and GNNs. Experiments across multiple datasets and models validate the proposed module.

LLM Enhancers for GNNs: An Analysis from the Perspective of Causal Mechanism Identification

The use of large language models (LLMs) as feature enhancers to optimize node representations, which are then used as inputs for graph neural networks (GNNs), has shown significant potential in graph representation learning. However, the fundamental properties of this approach remain underexplored. To address this issue, we propose conducting a more in-depth analysis of this issue based on the interchange intervention method. First, we construct a synthetic graph dataset with controllable causal relationships, enabling precise manipulation of semantic relationships and causal modeling to provide data for analysis. Using this dataset, we conduct interchange interventions to examine the deeper properties of LLM enhancers and GNNs, uncovering their underlying logic and internal mechanisms. Building on the analytical results, we design a plug-and-play optimization module to improve the information transfer between LLM enhancers and GNNs. Experiments across multiple datasets and models validate the proposed module.

ReactDance: Progressive-Granular Representation for Long-Term Coherent Reactive Dance Generation

Reactive dance generation (RDG) produces follower movements conditioned on guiding dancer and music while ensuring spatial coordination and temporal coherence. However, existing methods overemphasize global constraints and optimization, overlooking local information, such as fine-grained spatial interactions and localized temporal context. Therefore, we present ReactDance, a novel diffusion-based framework for high-fidelity RDG with long-term coherence and multi-scale controllability. Unlike existing methods that struggle with interaction fidelity, synchronization, and temporal consistency in duet synthesis, our approach introduces two key innovations: 1)Group Residual Finite Scalar Quantization (GRFSQ), a multi-scale disentangled motion representation that captures interaction semantics from coarse body rhythms to fine-grained joint dynamics, and 2)Blockwise Local Context (BLC), a sampling strategy eliminating error accumulation in long sequence generation via local block causal masking and periodic positional encoding. Built on the decoupled multi-scale GRFSQ representation, we implement a diffusion model withLayer-Decoupled Classifier-free Guidance (LDCFG), allowing granular control over motion semantics across scales. Extensive experiments on standard benchmarks demonstrate that ReactDance surpasses existing methods, achieving state-of-the-art performance.

TimeSoccer: An End-to-End Multimodal Large Language Model for Soccer Commentary Generation

Soccer is a globally popular sporting event, typically characterized by long matches and distinctive highlight moments. Recent advances in Multimodal Large Language Models (MLLMs) offer promising capabilities in temporal grounding and video understanding, soccer commentary generation often requires precise temporal localization and semantically rich descriptions over long-form video. However, existing soccer MLLMs often rely on the temporal a priori for caption generation, so they cannot process the soccer video end-to-end. While some traditional approaches follow a two-step paradigm that is complex and fails to capture the global context to achieve suboptimal performance. To solve the above issues, we present TimeSoccer, the first end-to-end soccer MLLM for Single-anchor Dense Video Captioning (SDVC) in full-match soccer videos. TimeSoccer jointly predicts timestamps and generates captions in a single pass, enabling global context modeling across 45-minute matches. To support long video understanding of soccer matches, we introduce MoFA-Select, a training-free, motion-aware frame compression module that adaptively selects representative frames via a coarse-to-fine strategy, and incorporates complementary training paradigms to strengthen the model's ability to handle long temporal sequences. Extensive experiments demonstrate that our TimeSoccer achieves State-of-The-Art (SoTA) performance on the SDVC task in an end-to-end form, generating high-quality commentary with accurate temporal alignment and strong semantic relevance.

VCR-Bench: A Comprehensive Evaluation Framework for Video Chain-of-Thought Reasoning

The advancement of Chain-of-Thought (CoT) reasoning has significantly enhanced the capabilities of large language models (LLMs) and large vision-language models (LVLMs). However, a rigorous evaluation framework for video CoT reasoning remains absent. Current video benchmarks fail to adequately assess the reasoning process and expose whether failures stem from deficiencies in perception or reasoning capabilities. Therefore, we introduce VCR-Bench, a novel benchmark designed to comprehensively evaluate LVLMs' Video Chain-of-Thought Reasoning capabilities. VCR-Bench comprises 859 videos spanning a variety of video content and durations, along with 1,034 high-quality question-answer pairs. Each pair is manually annotated with a stepwise CoT rationale, where every step is tagged to indicate its association with the perception or reasoning capabilities. Furthermore, we design seven distinct task dimensions and propose the CoT score to assess the entire CoT process based on the stepwise tagged CoT rationals. Extensive experiments on VCR-Bench highlight substantial limitations in current LVLMs. Even the top-performing model, o1, only achieves a 62.8% CoT score and an 56.7% accuracy, while most models score below 40%. Experiments show most models score lower on perception than reasoning steps, revealing LVLMs' key bottleneck in temporal-spatial information processing for complex video reasoning. A robust positive correlation between the CoT score and accuracy confirms the validity of our evaluation framework and underscores the critical role of CoT reasoning in solving complex video reasoning tasks. We hope VCR-Bench to serve as a standardized evaluation framework and expose the actual drawbacks in complex video reasoning task.

Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models

DeepSeek-R1-Zero has successfully demonstrated the emergence of reasoning capabilities in LLMs purely through Reinforcement Learning (RL). Inspired by this breakthrough, we explore how RL can be utilized to enhance the reasoning capability of MLLMs. However, direct training with RL struggles to activate complex reasoning capabilities such as questioning and reflection in MLLMs, due to the absence of substantial high-quality multimodal reasoning data. To address this issue, we propose the reasoning MLLM, Vision-R1, to improve multimodal reasoning capability. Specifically, we first construct a high-quality multimodal CoT dataset without human annotations by leveraging an existing MLLM and DeepSeek-R1 through modality bridging and data filtering to obtain a 200K multimodal CoT dataset, Vision-R1-cold dataset. It serves as cold-start initialization data for Vision-R1. To mitigate the optimization challenges caused by overthinking after cold start, we propose Progressive Thinking Suppression Training (PTST) strategy and employ Group Relative Policy Optimization (GRPO) with the hard formatting result reward function to gradually refine the model's ability to learn correct and complex reasoning processes on a 10K multimodal math dataset. Comprehensive experiments show our model achieves an average improvement of $\sim$6% across various multimodal math reasoning benchmarks. Vision-R1-7B achieves a 73.5% accuracy on the widely used MathVista benchmark, which is only 0.4% lower than the leading reasoning model, OpenAI O1. The datasets and code will be released in: https://github.com/Osilly/Vision-R1 .

LLaVA-RadZ: Can Multimodal Large Language Models Effectively Tackle Zero-shot Radiology Recognition?

Recently, multimodal large models (MLLMs) have demonstrated exceptional capabilities in visual understanding and reasoning across various vision-language tasks. However, MLLMs usually perform poorly in zero-shot medical disease recognition, as they do not fully exploit the captured features and available medical knowledge. To address this challenge, we propose LLaVA-RadZ, a simple yet effective framework for zero-shot medical disease recognition. Specifically, we design an end-to-end training strategy, termed Decoding-Side Feature Alignment Training (DFAT) to take advantage of the characteristics of the MLLM decoder architecture and incorporate modality-specific tokens tailored for different modalities, which effectively utilizes image and text representations and facilitates robust cross-modal alignment. Additionally, we introduce a Domain Knowledge Anchoring Module (DKAM) to exploit the intrinsic medical knowledge of large models, which mitigates the category semantic gap in image-text alignment. DKAM improves category-level alignment, allowing for accurate disease recognition. Extensive experiments on multiple benchmarks demonstrate that our LLaVA-RadZ significantly outperforms traditional MLLMs in zero-shot disease recognition and exhibits the state-of-the-art performance compared to the well-established and highly-optimized CLIP-based approaches.