InfoMerge is a novel, training-free method that significantly compresses visual tokens for Video-LLMs by estimating temporal redundancy and allocating tokens based on content richness, achieving high…

V-LynX is a framework that enhances Video LLMs by integrating new modalities into their existing token interface, achieving state-of-the-art performance across diverse video understanding tasks.

The paper analyzes token reduction for efficient unified VLM training, finding that while task-specific acceleration saves computation, it destroys the mutual performance gains achieved through joint…

The paper introduces Moment-Video, a new benchmark that diagnoses the ability of video MLLMs to understand brief, critical visual events, revealing that current models struggle significantly with temp…

肖代替了视觉令牌的永久删除，通过可恢复的路由来改进视觉语言模型的性能

VideoMLA introduces a novel Multi-Head Latent Attention (MLA) mechanism that replaces per-head KV caches with a shared low-rank content latent, significantly reducing memory and improving throughput f…

The paper introduces MLLM-Microscope, a system that analyzes the internal structure of multimodal large language models (MLLMs), finding that modality fusion significantly impacts the linearity and di…

CIVIC is a path-consistent compact visual inference framework that achieves genuine hardware efficiency in Vision-Language Models by maintaining contiguous sequence representations across all inferenc…

OccamToken introduces a training-free, adaptive token pruning framework that replaces fixed token budgets with relative evidence testing against a register-based reference, significantly improving VLM…

The paper introduces PaSBench-Video, a comprehensive streaming video benchmark designed to rigorously test multimodal LLMs' ability to issue proactive safety warnings, finding that current models stru…

The paper proposes PrefixMem, a dedicated encoder for Semantic IDs (SIDs), demonstrating that structured, prefix-conditioned representations significantly improve the accuracy and recall of generative…

The paper proposes VLM3, a simple, scalable method that demonstrates standard Vision Language Models (VLMs) can natively learn 3D understanding by focusing on architectural simplicity and specific dat…

The paper demonstrates that specialized coding agents, using only text and image access within a sandbox, can effectively solve complex omnimodal tasks, often outperforming state-of-the-art native omn…

The paper introduces Multi-Clip Video (MCV) SafetyBench, a dataset demonstrating that the vulnerability of Multimodal Large Language Models (MLLMs) to jailbreaking increases with the diversity and num…

This paper presents a black-box membership inference attack (MIA) against Video Large Language Models (VideoLLMs), demonstrating that they are vulnerable by analyzing generation behavior across varyin…

The paper proposes a unified framework that decouples long-video reasoning into semantic and visual evidence, significantly improving performance on the HD-EPIC VQA Challenge.