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…

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…

AdaCodec introduces a predictive visual coding scheme for video MLLMs, significantly improving efficiency and performance by transmitting only inter-frame changes and full reference frames when necess…

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

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…

The paper proposes a training-free framework, Visual Representation-Guided Video-LLM Reasoning, to perform composed video retrieval by using visual examples and text instructions, achieving strong per…

Lumos-Nexus is a training-efficient framework that enhances video generation quality by progressively bridging generation from a lightweight model to a high-fidelity generator in a shared latent space…

Zamba2-VL is a new suite of vision-language models built on the Zamba2 hybrid architecture, achieving state-of-the-art performance and significantly improved inference efficiency compared to leading T…

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…

This paper systematically analyzes how different architectural components of Large Vision-Language Models (LVLMs) contribute to hallucination robustness, finding that joint enhancement of visual fidel…

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…

PARCEL introduces a novel visual tokenization architecture that combines spatial pooling anchors with conditioned elastic queries, efficiently reducing the computational cost of large Vision-Language…

VidPrism introduces a novel heterogeneous Mixture-of-Experts framework that specializes temporal processing by dividing labor among experts, achieving state-of-the-art performance in image-to-video tr…

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

The paper proposes a novel global multi-modal alignment framework to robustly learn video representations from noisy and complementary sensor data, significantly improving driver distraction detection…