Saved in:
Bibliographic Details
Main Author: Li, Chenjun
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.04676
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910199987568640
author Li, Chenjun
author_facet Li, Chenjun
contents Multi-image reasoning remains a significant challenge for vision-language models (VLMs). We investigate a previously overlooked phenomenon: during chain-of-thought (CoT) generation, the text-to-image (T2I) attention of reasoning VLMs exhibits diffuse "pulses": sporadic and unfocused attention patterns that fail to concentrate on task-relevant images. We further reveal a systematic positional bias in attention allocation across images. Motivated by these observations, we propose PulseFocus, a training-free, inference-time method that structures CoT reasoning into interleaved plan/focus blocks with soft attention gating. By forcing the model to explicitly plan which image to examine and then gating decode-time attention to the referenced image, PulseFocus sharpens attention focus and yields consistent improvements on multi-image benchmarks like BLINK benchmark (+3.7%) and MuirBench (+1.07%).
format Preprint
id arxiv_https___arxiv_org_abs_2603_04676
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Decoding the Pulse of Reasoning VLMs in Multi-Image Understanding Tasks
Li, Chenjun
Computer Vision and Pattern Recognition
Artificial Intelligence
Multi-image reasoning remains a significant challenge for vision-language models (VLMs). We investigate a previously overlooked phenomenon: during chain-of-thought (CoT) generation, the text-to-image (T2I) attention of reasoning VLMs exhibits diffuse "pulses": sporadic and unfocused attention patterns that fail to concentrate on task-relevant images. We further reveal a systematic positional bias in attention allocation across images. Motivated by these observations, we propose PulseFocus, a training-free, inference-time method that structures CoT reasoning into interleaved plan/focus blocks with soft attention gating. By forcing the model to explicitly plan which image to examine and then gating decode-time attention to the referenced image, PulseFocus sharpens attention focus and yields consistent improvements on multi-image benchmarks like BLINK benchmark (+3.7%) and MuirBench (+1.07%).
title Decoding the Pulse of Reasoning VLMs in Multi-Image Understanding Tasks
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2603.04676