← 研踪 PaperTrailarXiv:2611.20936 · v1
ReAccess: Look Once, Stay Focused with Efficient Test Time Memory
首次提交 2026-11-16 19:46
版本更新 2026-11-16 19:46
雷达观测 2026-11-17 17:11
摘要
Despite recent advances, vision-language models (VLMs) still struggle with fine-grained visual understanding, often failing to effectively utilize critical evidence already encoded in their visual representations. Existing approaches address this limitation through repeated visual processing or explicit reasoning, introducing substantial computational overhead. In this paper, we propose ReAccess, a lightweight framework that preserves and reuses question-relevant visual evidence through compact associative memory. Constructed during a single visual prefill, the memory supports progressive evidence retrieval that steers both attention queries and hidden representations, enabling effective visual grounding without repeated image encoding or explicit reasoning traces. Extensive experiments demonstrate consistent improvements in both general visual understanding and fine-grained perception, while achieving substantially faster inference than existing reasoning-intensive approaches, demonstrating a promising path to efficient multimodal reasoning.
作者备注:Project Page: https://ryu1ion.github.io/reaccess/
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-16 19:46 · 最新观测版本
ReAccess: Look Once, Stay Focused with Efficient Test Time Memory
Despite recent advances, vision-language models (VLMs) still struggle with fine-grained visual understanding, often failing to effectively utilize critical evidence already encoded in their visual representations. Existing approaches address this limitation through repeated visual processing or explicit reasoning, introducing substantial computational overhead. In this paper, we propose ReAccess, a lightweight framework that preserves and reuses question-relevant visual evidence through compact associative memory. Constructed during a single visual prefill, the memory supports progressive evidence retrieval that steers both attention queries and hidden representations, enabling effective visual grounding without repeated image encoding or explicit reasoning traces. Extensive experiments demonstrate consistent improvements in both general visual understanding and fine-grained perception, while achieving substantially faster inference than existing reasoning-intensive approaches, demonstrating a promising path to efficient multimodal reasoning.
查看此版本解读 →查看此版本原文 ↗