论文检索

Zining Liu, Yunhai Hu, Tianhua Xia, BO Bao, Eric Sather, Vithursan Thangarasa, Sai Qian Zhang

Speculative decoding (SD) has proven to be an effective technique for accelerating autoregressive generation in large language models (LLMs), however its application to vision-language models (VLMs) remains relatively unexplored. We propose DREAM-S, a novel SD framework designed specifically for fast and efficient decoding in VLMs. DREAM-S leverages a neural architecture search (NAS) framework with target-aware supernet training to automatically identify both the optimal interaction strategy between the draft and target models, and the most suitable draft model architecture for the underlying hardware implementation platform. DREAM-S additionally incorporates adaptive intermediate feature distillation, guided by attention entropy, to enable efficient draft training. Experiments on a range of well-established VLMs show that DREAM-S achieves up to a 3.85\times speedup compared to standard decoding approaches and significantly outperforms existing SD baselines.

Duy Nguyen, Archiki Prasad, Elias Stengel-Eskin, Mohit Bansal

Inference-time steering provides a lightweight alternative to fine-tuning large language models (LLMs) and vision-language models (VLMs) by modifying model activations without updating weights. However, existing methods often rely on a global intervention vector, overlook token-level causal influence, and underutilize model logits, especially in multimodal settings where visual and textual inputs contribute unevenly. We propose GrAInS, a contrastive, gradient-based approach that leverages Integrated Gradients to identify top-k influential tokens and construct directional steering vectors based on their contribution to preferred over dispreferred outputs. These vectors guide activation intervention at each layer, preserving the representational scale. GrAInS outperforms fine-tuning and prior steering methods on both LLM and VLM tasks: improving TruthfulQA accuracy by 13.22% (Llama-3.1-8B), reducing MMHal-Bench hallucinations from 0.624 to 0.514 (LLaVA-1.6-7B), and increasing SPA-VL alignment by 8.11%, all without degrading fluency or general capabilities.

Rohit Sinha, Aditya Sanjiv Kanade, Sai Srinivas Kancheti, Vineeth N. Balasubramanian, Tanuja Ganu

Multimodal large language models (MLLMs) have achieved impressive progress on vision language benchmarks, yet their capacity for cognitive and psychological reasoning remains largely unexplored. We introduce Mind’s Eye, a multiple-choice benchmark of eight visuo-cognitive tasks inspired by classic human intelligence tests and organized under a novel A–R–T taxonomy: Abstraction, Relation, and Transformation. The tasks probe core processes of fluid intelligence such as pattern induction, analogical Relation mapping, and mental Transformation. We evaluate a diverse suite of closed-source and open-source MLLMs and compare their performance with human participants. Humans achieve 80% accuracy, while top performing MLLMs remain below 50%. Error analysis reveals failures in (i) visual attention allocation, (ii) internal perceptual manipulation, (iii) over reliance on domain priors, and (iv) weak abstraction of underlying visual concepts. Our findings suggest that current MLLMs exhibit limited fluid reasoning and visuo-cognitive integration compared with human participants, highlighting the need for cognitively grounded evaluation frameworks like Mind’s Eye.

Qiguang Chen, Chengyu Luan, Jiajun Wu, Qiming Yu, Yi Yang, Yizhuo Li, Jingqi Tong, Xiachong Feng, Libo Qin, Wanxiang Che

Large vision-language models (LVLMs) have made substantial advances in reasoning tasks at the Olympiad level. Nevertheless, current Olympiad-level multimodal reasoning benchmarks for these models often emphasize single-image analysis and fail to exploit contextual information across multiple images. We present OMIBench, a benchmark designed to evaluate Olympiad-level reasoning when the required evidence is distributed over multiple images. It contains problems from biology, chemistry, mathematics, and physics Olympiads, together with manually annotated rationales and evaluation protocols for both exact and semantic answer matching. Across extensive experiments on OMIBench, we observe meaningful performance gaps in existing models. Even the strongest LVLMs, such as Gemini-3-Pro, attain only about 50% on the benchmark. These results position OMIBench as a focused resources for studying and improving multi-image reasoning in LVLMs.

Wenyi Xiao, Xinchi XU, Leilei Gan

Large Vision Language Models (LVLMs) achieve strong multimodal reasoning but frequently exhibit hallucinations and incorrect responses with high certainty, which hinders their usage in high-stakes domains. Existing verbalized confidence calibration methods, largely developed for text-only LLMs, typically optimize a single holistic confidence score using binary answer-level correctness. This design is mismatched to LVLMs: an incorrect prediction may arise from perceptual failures or from reasoning errors given correct perception, and a single confidence conflates these sources while visual uncertainty is often dominated by language priors. To address these issues, we propose VL-Calibration, a reinforcement learning framework that explicitly decouples confidence into visual and reasoning confidence. To supervise visual confidence without ground-truth perception labels, we introduce an intrinsic visual certainty estimation that combines (i) visual grounding measured by KL-divergence under image perturbations and (ii) internal certainty measured by token entropy. We further propose token-level advantage reweighting to focus optimization on tokens based on visual certainty, suppressing ungrounded hallucinations while preserving valid perception. Experiments on thirteen benchmarks show that VL-Calibration effectively improves calibration while boosting visual reasoning accuracy, and it generalizes to out-of-distribution benchmarks across model scales and architectures.

Hyosik Moon, Eldan Cohen

Behavior trees provide a transparent and modular structure for encoding expert-designed policies, enabling interpretable decision-making in complex tasks. Yet, applying behavior trees to high-dimensional perceptual inputs such as images or language is challenging as defining symbolic predicates over raw perceptual data is non-trivial. While state-of-the-art large multimodal models (such as vision-language models) can overcome this issue by utilizing natural language queries over perceptual inputs, they incur high computational cost, making them unsuitable for many applications. Imitation learning offers a way to distill these expert models into compact models, though it requires extensive supervision. In contrast, reinforcement learning reduces the need for costly supervision but risks misalignment of condition nodes with their intended semantics as well as poor credit assignment. To address these challenges, we introduce CERL (Condition-node Expert-regularized Reinforcement Learning), a framework that leverages expert-regularized reinforcement learning to preserve semantic faithfulness, while employing a factorized policy that aggregates sequential condition-node decisions into a single decision unit to alleviate credit assignment challenges. Experiments across seven tasks from the GymCards, FrozenLake, and BabyAIText suites demonstrate that our framework outperforms pure imitation learning or reinforcement learning baselines, retains strong agreement with expert decisions, and achieves substantial gains in inference speed and model size over expert models. Our implementation is available in https://github.com/HyosikMoon/CERL.

Onur Keleş, Asli Ozyurek, Gerardo Ortega, Kadir Gökgöz, Esam Ghaleb

Iconicity, the resemblance between linguistic form and meaning, is pervasive in sign languages, offering a natural testbed for visual grounding in vision–language models (VLMs). We introduce the Visual Iconicity Challenge, a video-based benchmark that adapts psycholinguistic measures to evaluate VLMs on three tasks: (i) phonological sign-form prediction, (ii) transparency (inferring meaning from visual form), and (iii) graded iconicity ratings. We assess 17 state-of-the-art VLMs in zero- and few-shot settings on Sign Language of the Netherlands and compare them to human baselines. VLMs mirror human phonological difficulty patterns (e.g., handshape harder than location) and achieve moderate to strong alignment with human iconicity ratings. However, they still fail to infer lexical meaning from visual form alone and show a systematic object-based bias that inverts the human preference for action-based signs. Crucially, models with stronger phonological form prediction correlate better with human iconicity judgments, indicating shared sensitivity to visually grounded structure. Our findings validate these diagnostic tasks, show that explicit reasoning narrows the open-to-closed-model calibration gap, and motivate human-centric signals for modelling iconicity in multimodal models.

Haodong Chen, Qiang Huang, Jiaqi Zhao, Qiuping Jiang, Xiaojun Chang, Jun Yu

Vision-Language Models (VLMs) are increasingly deployed in socially consequential settings, raising concerns about social bias driven by demographic cues. A central challenge in measuring such social bias is attribution under visual confounding: real-world images entangle race and gender with correlated factors such as background and clothing, obscuring attribution. We propose a **face-only counterfactual evaluation paradigm** that isolates demographic effects while preserving real-image realism. Starting from real photographs, we generate counterfactual variants by editing only facial attributes related to race and gender, keeping all other visual factors fixed. Based on this paradigm, we construct **FOCUS**, a dataset of 480 scene-matched counterfactual images across six occupations and ten demographic groups, and propose **REFLECT,** a benchmark comprising three decision-oriented tasks: two-alternative forced choice, multiple-choice socioeconomic inference, and numeric salary recommendation. Experiments on five state-of-the-art VLMs reveal that demographic disparities persist under strict visual control and vary substantially across task formulations. These findings underscore the necessity of controlled, counterfactual audits and highlight task design as a critical factor in evaluating social bias in multimodal models.

Manyi Zhang, Ji-Fu Li, Zhongao Sun, Haoli Bai, Hui-Ling Zhen, Zhenhua Dong, Xianzhi Yu

Microscaling Floating-Point (MXFP) has emerged as a promising low-precision format for large language models (LLMs). Despite various post-training quantization (PTQ) algorithms being proposed, they mostly focus on integer quantization, while their applicability and behavior under MXFP formats remain largely unexplored. To address this gap, this work conducts a systematic investigation of PTQ under MXFP formats, encompassing over 7 PTQ algorithms, 15 evaluation benchmarks, and 3 LLM families. The key findings include: 1) MXFP8 consistently achieves near-lossless performance, while MXFP4 introduces substantial accuracy degradation and remains challenging; 2) PTQ effectiveness under MXFP depends strongly on format compatibility, with some algorithmic paradigms being consistently more effective than others; 3) PTQ performance exhibits highly consistent trends across model families and modalities, in particular, quantization sensitivity is dominated by the language model rather than the vision encoder in multimodal LLMs; 4) The scaling factor of quantization is a critical error source in MXFP4, and a simple pre-scale optimization strategy can significantly mitigate its impact. Together, these results provide practical guidance on adapting existing PTQ methods to MXFP quantization.

Xiyao Dong, Guangsheng Cheng, YiLong Chen, Xiaojin Zhang, Kun He

Recent advances in Multimodal Large Reasoning Models (MLRMs) have enabled explicit chain-of-thought inference across vision and language, substantially improving performance on complex reasoning tasks. Despite these gains, the reasoning process introduces a subtle yet critical vulnerability. We identify an underexplored multimodal safety failure mode in which harmful objectives are embedded within ostensibly benign contexts, leading models to over-prioritize narrative coherence during reasoning. We term this phenomenon Safety Context Amnesia (SCA), wherein models correctly perceive risk-relevant visual cues but fail to enforce safety constraints as the reasoning process becomes dominated by contextual alignment. To mitigate SCA, we propose Intent-Guided Safety Reasoning (IGSR), an inference-time defense that operates without modifying target model parameters. IGSR employs a Perception Decoupler to extract objective visual evidence into a structured intent output, followed by a Cognitive Arbiter that enforces explicit safety constraints prior to generation. Extensive experiments across multiple multimodal safety benchmarks demonstrate that IGSR improves defense success rates by over 62% compared to baselines, while largely preserving task utility. These results highlight the critical role of structured, intent-aware reasoning in achieving robust safety reasoning for multimodal reasoning models.

Zhengpeng Shi, Yanpeng Zhao, Jianqun Zhou, Yuxuan Wang, Qinrong Cui, Wei Bi, Song-Chun Zhu, Bo Zhao, Zilong Zheng

AI models capable of comprehending humor hold real-world promise—for example, enhancing engagement in human-machine interactions. To gauge and diagnose the capacity of multimodal large language models (MLLMs) for humor understanding, we introduce v-HUB, a novel video humor understanding benchmark. v-HUB comprises a curated collection of non-verbal short videos, reflecting real-world scenarios where humor can be appreciated purely through visual cues. We pair each video clip with rich annotations to support a variety of evaluation tasks and analyses, including a novel study of environmental sound that can enhance humor. To broaden its applicability, we construct an open-ended QA task, making v-HUB readily integrable into existing video understanding task suites. We evaluate a diverse set of MLLMs, from specialized Video-LLMs to versatile OmniLLMs that can natively process audio, covering both open-source and proprietary domains. The experimental results expose the difficulties MLLMs face in comprehending humor from visual cues alone. Our findings also demonstrate that incorporating audio helps with video humor understanding, highlighting the promise of integrating richer modalities for complex video understanding tasks.

Shuliang Liu, Songbo Yang, Dong Fang, Sihang Jia, Yuqi Tang, Lingfeng Su, Ruoshui Peng, Yibo Yan, Xin Zou, Xuming Hu

Object hallucination critically undermines the reliability of Multimodal Large Language Models (MLLMs), often stemming from a fundamental failure in cognitive introspection—where models blindly trust linguistic priors over specific visual evidence. Existing mitigations remain limited: contrastive decoding approaches operate superficially without rectifying internal semantic misalignments, while current latent steering methods rely on static vectors that lack instance-specific precision. We introduce Vision-Language Introspection (VLI), a training-free inference framework that simulates a metacognitive self-correction process. VLI first performs Attributive Introspection to diagnose hallucination risks via probabilistic conflict detection and localize the causal visual anchors. It then employs Interpretable Bi-Causal Steering to actively modulate the inference process, dynamically isolating visual evidence from background noise while neutralizing blind confidence through adaptive calibration. VLI achieves state-of-the-art performance on advanced models, reducing object hallucination rates by 12.67% on MMHal-Bench and improving accuracy by 5.8% on POPE.

Yangneng Chen, Junlin Li, Weijun Yao, Xilai Ma, Guodong DU, Wenya Wang, Jing Li

Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal tasks, yet their reliability is persistently undermined by hallucinations—generating text that contradicts visual input. Recent studies often attribute these errors to inadequate visual attention. In this work, we analyze the attention mechanisms via the logit lens, uncovering a distinct anomaly we term **Vocabulary Hijacking**. We discover that specific visual tokens, defined as **Inert Tokens**, disproportionately attract attention. Crucially, when their intermediate hidden states are projected into the vocabulary space, they consistently decode to a fixed set of unrelated words (termed **Hijacking Anchors**) across layers, revealing a rigid semantic collapse. Leveraging this semantic rigidity, we propose **Hijacking Anchor-Based Identification (HABI)**, a robust strategy to accurately localize these Inert Tokens. To quantify the impact of this phenomenon, we introduce the **Non-Hijacked Visual Attention Ratio (NHAR)**, a novel metric designed to identify attention heads that remain resilient to hijacking and are critical for factual accuracy. Building on these insights, we propose **Hijacking-Aware Visual Attention Enhancement (HAVAE)**, a training-free intervention that selectively strengthens the focus of these identified heads on salient visual content. Extensive experiments across multiple benchmarks demonstrate that HAVAE significantly mitigates hallucinations with **no additional computational overhead**, while preserving the model’s general capabilities.

Akash Ghosh, Subhadip Baidya, Sriparna Saha, Xiuying Chen

Vision–Language Models (VLMs) are increasingly used in clinical diagnostics, yet their robustness to adversarial attacks remains largely unexplored, posing serious risks. Existing medical attacks focus on secondary objectives such as model stealing or adversarial fine-tuning, while transferable attacks from natural images introduce visible distortions that clinicians can easily detect. To address this, we propose MedFocusLeak, a highly transferable black-box multimodal attack that induces incorrect yet clinically plausible diagnoses while keeping perturbations imperceptible. The method injects coordinated perturbations into non-diagnostic background regions and employs an attention-distraction mechanism to shift the model’s focus away from pathological areas. Extensive evaluations across six medical imaging modalities show that MedFocusLeak achieves state-of-the-art performance, generating misleading yet realistic diagnostic outputs across diverse VLMs. We further introduce a unified evaluation framework with novel metrics that jointly capture attack success and image fidelity, revealing a critical weakness in the reasoning capabilities of modern clinical VLMs.

Haosen Wang, Jing Xiao, Chaochao Du, Xiaowang Zhang, Zhiyong Feng

Document Visual Question Answering (DocVQA) aims to generate answers by jointly understanding the textual, layout, and visual elements within document images. Although end-to-end vision-based generative methods have reduced dependency on OCR, they still struggle to achieve precise evidence localization when page semantics are complex and highly similar. However, existing research lacks an in-depth theoretical analysis of the question-driven semantic representation space, failing to fundamentally address the distinguishability problem among semantically similar pages. To fill this theoretical gap, we propose and prove that, given a specific question, each page possesses a unique semantic representation, and there exists a bijective mapping between the page and its unique semantics. Based on this theoretical foundation, we introduce the Flow-Based Page Unique Semantic Mapping Architecture (FUMA), which reconstructs evidence localization from similarity-based retrieval into precise selection on unique semantics. FUMA employs fine-grained cross-modal attention to extract discriminative cues and utilizes flow-based reversible transformations with likelihood regularization to learn bijective mappings, ensuring that each page obtains a unique semantic representation. Moreover, a multi-expert collaboration mechanism complementarily models fine-grained multimodal information within each page, achieving robust answer generation. Experimental results demonstrate that FUMA significantly outperforms existing methods in both evidence localization and answer generation.

Jianing Zhang, Runan Li, Honglin Pang, Ding Xia, Zhou Zhu, Qian Zhang, Chuntao Li, Xi Yang

Deciphering ancient Chinese Oracle Bone Script (OBS) is a challenging task that offers insights into the beliefs, systems, and culture of the ancient era. Existing approaches treat decipherment as a closed-set image recognition problem, which fails to bridge the “interpretation gap”: while individual characters are often unique and rare, they are composed of a limited set of recurring, pictographic components that carry transferable semantic meanings. To leverage this structural logic, we propose an agent-driven Vision-Language Model (VLM) framework that integrates a VLM for precise visual grounding with an LLM-based agent to automate a reasoning chain of component identification, graph-based knowledge retrieval, and relationship inference for linguistically accurate interpretation. To support this, we also introduce OB-Radix, an expert-annotated dataset providing structural and semantic data absent from prior corpora, comprising 1,022 character images (934 unique characters) and 1,853 fine-grained component images across 478 distinct components with verified explanations. By evaluating our system across three benchmarks of different tasks, we demonstrate that our framework yields more detailed and precise decipherments compared to baseline methods.

Zoher Kachwala, Bao Tran Truong, Rasika Muralidharan, Haewoon Kwak, Jisun An, Filippo Menczer

Social media are shifting towards pluralism — community-governed platforms where groups define their own norms. What violates rules in one community may be perfectly acceptable in another. Can AI models help moderate such pluralistic communities? We formalize the task as a multiple-choice problem, mirroring how human moderators operate in the real world: given a comment and its surrounding context, identify which specific rule, if any, is violated. We introduce PluRule, a multimodal, multilingual benchmark for detecting 13,371 rule violations across 1,989 Reddit communities spanning 2,885 rules in 9 languages. Using this benchmark, we show that state-of-the-art vision-language models struggle significantly: even GPT-5.2 with high reasoning performs only slightly better than a trivial baseline. We also find that bigger models and increased context provide marginal gains, and universal rules like civility and self-promotion are easier to detect. Our results show that moderation of pluralistic communities on social media is a fundamental challenge for language models. Our code and benchmark are publicly available.

Yucheng Zhou, Dubing Chen, Huan Zheng, Jianbing Shen

Recent advances in text-to-image (T2I) generation have enabled visually coherent image synthesis from descriptions, but generating images containing multiple given subjects remains challenging. As the number of reference identities increases, existing methods often suffer from subject missing and semantic drift. To address this problem, we propose MUSIC, the first MLLM specifically designed for MUlti-Subject In-Context image generation. To overcome the data scarcity, we introduce an automatic and scalable data generation pipeline that eliminates the need for manual annotation. Furthermore, we enhance the model’s understanding of multi-subject semantic relationships through a vision chain-of-thought (CoT) mechanism, guiding step-by-step reasoning from subject images to semantics and generation. To mitigate identity entanglement and manage visual complexity, we develop a novel semantics-driven spatial layout planning method and demonstrate its test-time scalability. By incorporating complex subject images during training, we improve the model’s capacity for chained reasoning. In addition, we curate MSIC, a new benchmark tailored for multi-subject in-context generation. Experimental results demonstrate that MUSIC significantly surpasses other methods in both multi- and single-subject scenarios.

Yoonji Kim, Jieun Kim, Yujin Jeong, Sung-Bae Cho

Consistent reasoning about 3D spatial relations across changing viewpoints is fundamental for Embodied AI agents operating in dynamic environments. While Large Vision-Language Models (LVLMs) have advanced multimodal perception, their ability to maintain spatial consistency across diverse perspectives remains underexplored. Existing benchmarks primarily assess spatial capabilities from a static, single-view, and egocentric perspective, failing to capture the dynamic nature of real-world spatial cognition.To address this gap, we introduce SCOPE (Spatial COnsistency across PErspectives and Viewpoints), a comprehensive benchmark designed to rigorously diagnose spatial reasoning capabilities. Grounded in human cognitive theories of dual spatial representations, SCOPE discretizes the 360^\circ field into multiview scenarios to systematically evaluate both allocentric and egocentric reasoning capabilities. Our dataset comprises 20.1K spatial VQA pairs derived from high-quality 3D environments. Through an extensive evaluation of 26 state-of-the-art LVLMs, we identify two fundamental limitations that prevent consistent spatial understanding across viewpoints.We hope SCOPE facilitates the diagnosis of spatial reasoning, serving as a stepping stone toward reliable embodied action.

Wei Li, Zhen Huang, Xinmei Tian

Contrastively trained vision-language models like CLIP, have made remarkable progress in learning joint image-text representations, but still face challenges in compositional understanding. They often exhibit a “bag-of-words” behavior—struggling to capture the object relations, attribute-object bindings, and word order dependencies. This limitation arises not only from the reliance on global, single-vector representations for optimization, but also from the insufficient exploitation and modeling of the rich compositional information inherently present in paired image text data. In this work, we propose **MACCO** (**MA**sked **C**ompositional **C**oncept M**O**deling), a framework that masks compositional concepts in one modality and reconstructs them conditioned on the full contextual information from the other, enabling the model to capture and align cross-modal compositional structures more effectively. To facilitate this process, we introduce two auxiliary objectives that jointly align and regularize masked features both inter-modally and intra-modally. Extensive experiments on five compositional benchmarks, along with in-depth analyses, demonstrate that our approach not only significantly enhances compositionality in VLMs but also improves their ability to capture syntactic structure and linguistic information. Additionally, the improved compositionality also benefits text-to-image generation and multimodal large language model.