Cover Image

Can Retrieval Heads See Images? Multimodal Retrieval Heads in Long-Context Vision-Language Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Li, Aaron Branson Cigres, Wang, Zhaowei, Zhao, Yu, Du, Yiming, Li, Haobo, Ren, Xiyu, Wong, Ginny, See, Simon, Luo, Lishu, Duan, Haodong, Minervini, Pasquale, Song, Yangqiu
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

MMLongBench: Benchmarking Long-Context Vision-Language Models Effectively and Thoroughly
by: Wang, Zhaowei, et al.
Published: (2025)

MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language Models
by: Ren, Xiyu, et al.
Published: (2026)

Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context
by: Wang, Zhaowei, et al.
Published: (2026)

$\mathbb{R}^{2k}$ is Theoretically Large Enough for Embedding-based Top-$k$ Retrieval
by: Wang, Zihao, et al.
Published: (2026)

DeCoRe: Decoding by Contrasting Retrieval Heads to Mitigate Hallucinations
by: Gema, Aryo Pradipta, et al.
Published: (2024)

Retrieval Head Mechanistically Explains Long-Context Factuality
by: Wu, Wenhao, et al.
Published: (2024)

VERA: Identifying and Leveraging Visual Evidence Retrieval Heads in Long-Context Understanding
by: Pei, Rongcan, et al.
Published: (2026)

Evaluating and Safeguarding the Adversarial Robustness of Retrieval-Based In-Context Learning
by: Yu, Simon, et al.
Published: (2024)

Retrieval Heads are Dynamic
by: Lin, Yuping, et al.
Published: (2026)

Answerability in Retrieval-Augmented Open-Domain Question Answering
by: Abdumalikov, Rustam, et al.
Published: (2024)

ZigzagAttention: Efficient Long-Context Inference with Exclusive Retrieval and Streaming Heads
by: Liu, Zhuorui, et al.
Published: (2025)

From Interpretability to Performance: Optimizing Retrieval Heads for Long-Context Language Models
by: Ma, Youmi, et al.
Published: (2026)

Query-Focused Retrieval Heads Improve Long-Context Reasoning and Re-ranking
by: Zhang, Wuwei, et al.
Published: (2025)

DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
by: Xiao, Guangxuan, et al.
Published: (2024)

Understanding Synthetic Context Extension via Retrieval Heads
by: Zhao, Xinyu, et al.
Published: (2024)

The Curse of CoT: On the Limitations of Chain-of-Thought in In-Context Learning
by: Zheng, Tianshi, et al.
Published: (2025)

The Atlas of In-Context Learning: How Attention Heads Shape In-Context Retrieval Augmentation
by: Kahardipraja, Patrick, et al.
Published: (2025)

AbsInstruct: Eliciting Abstraction Ability from LLMs through Explanation Tuning with Plausibility Estimation
by: Wang, Zhaowei, et al.
Published: (2024)

LongHeads: Multi-Head Attention is Secretly a Long Context Processor
by: Lu, Yi, et al.
Published: (2024)

ComparisonQA: Evaluating Factuality Robustness of LLMs Through Knowledge Frequency Control and Uncertainty
by: Zong, Qing, et al.
Published: (2024)

Contrastive Retrieval Heads Improve Attention-Based Re-Ranking
by: Tran, Linh, et al.
Published: (2025)

PosterSum: A Multimodal Benchmark for Scientific Poster Summarization
by: Saxena, Rohit, et al.
Published: (2025)

LogiDynamics: Unraveling the Dynamics of Inductive, Abductive and Deductive Logical Inferences in LLM Reasoning
by: Zheng, Tianshi, et al.
Published: (2025)

A Parametric Memory Head for Continual Generative Retrieval
by: Mekonnen, Kidist Amde, et al.
Published: (2026)

Headings, Subject. See SUBJECT HEADINGS
by: Woods, William Edward
Published: (1972)

VLM-RobustBench: A Comprehensive Benchmark for Robustness of Vision-Language Models
by: Saxena, Rohit, et al.
Published: (2026)

XL-HeadTags: Leveraging Multimodal Retrieval Augmentation for the Multilingual Generation of News Headlines and Tags
by: Shohan, Faisal Tareque, et al.
Published: (2024)

ILRe: Intermediate Layer Retrieval for Context Compression in Causal Language Models
by: Liang, Manlai, et al.
Published: (2025)

The Rotary Position Embedding May Cause Dimension Inefficiency in Attention Heads for Long-Distance Retrieval
by: Chiang, Ting-Rui, et al.
Published: (2025)

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation
by: Lin, Xiaolin, et al.
Published: (2025)

Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning
by: Fu, Yu, et al.
Published: (2024)

Do Composed Image Retrieval Benchmarks Require Multimodal Composition?
by: Attimonelli, Matteo, et al.
Published: (2026)

Lost in Time: Clock and Calendar Understanding Challenges in Multimodal LLMs
by: Saxena, Rohit, et al.
Published: (2025)

Persona Knowledge-Aligned Prompt Tuning Method for Online Debate
by: Chan, Chunkit, et al.
Published: (2024)

Getting Sick After Seeing a Doctor? Diagnosing and Mitigating Knowledge Conflicts in Event Temporal Reasoning
by: Fang, Tianqing, et al.
Published: (2023)

MLDocRAG: Multimodal Long-Context Document Retrieval Augmented Generation
by: Zhang, Yongyue, et al.
Published: (2026)

LoVR: A Benchmark for Long Video Retrieval in Multimodal Contexts
by: Cai, Qifeng, et al.
Published: (2025)

UniHead: Unifying Multi-Perception for Detection Heads
by: Zhou, Hantao, et al.
Published: (2023)

Can Long-Context Language Models Subsume Retrieval, RAG, SQL, and More?
by: Lee, Jinhyuk, et al.
Published: (2024)

LongEmbed: Extending Embedding Models for Long Context Retrieval
by: Zhu, Dawei, et al.
Published: (2024)