Focus: A Streaming Concentration Architecture for Efficient Vision-Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wei, Chiyue, Guo, Cong, Zhang, Junyao, Shan, Haoxuan, Xu, Yifan, Zhang, Ziyue, Liu, Yudong, Wang, Qinsi, Zhou, Changchun, Li, Hai "Helen", Chen, Yiran
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909965627686912
author Wei, Chiyue
Guo, Cong
Zhang, Junyao
Shan, Haoxuan
Xu, Yifan
Zhang, Ziyue
Liu, Yudong
Wang, Qinsi
Zhou, Changchun
Li, Hai "Helen"
Chen, Yiran
author_facet Wei, Chiyue
Guo, Cong
Zhang, Junyao
Shan, Haoxuan
Xu, Yifan
Zhang, Ziyue
Liu, Yudong
Wang, Qinsi
Zhou, Changchun
Li, Hai "Helen"
Chen, Yiran
contents Vision-Language Models (VLMs) have demonstrated strong performance on tasks such as video captioning and visual question answering. However, their growing scale and video-level inputs lead to significant computational and memory overhead, posing challenges for real-time deployment on hardware accelerators. While prior work attempts to reduce redundancy via token pruning or merging, these methods typically operate at coarse granularity and incur high runtime overhead due to global token-level operations. In this study, we propose Focus, a Streaming Concentration Architecture that efficiently accelerates VLM inference through progressive, fine-grained redundancy elimination. Focus introduces a multilevel concentration paradigm that hierarchically compresses vision-language inputs at three levels: (1) semantic-guided token pruning based on textual prompts, (2) spatial-temporal block-level concentration using localized comparisons, and (3) vector-level redundancy removal via motion-aware matching. All concentration steps are tightly co-designed with the architecture to support streaming-friendly, on-chip execution. Focus leverages GEMM tiling, convolution-style layout, and cross-modal attention to minimize off-chip access while enabling high throughput. Implemented as a modular unit within a systolic-array accelerator, Focus achieves a 2.4x speedup and 3.3x reduction in energy, significantly outperforming state-of-the-art accelerators in both performance and energy efficiency. Full-stack implementation of Focus is open-sourced at https://github.com/dubcyfor3/Focus.
format Preprint
id arxiv_https___arxiv_org_abs_2512_14661
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Focus: A Streaming Concentration Architecture for Efficient Vision-Language Models
Wei, Chiyue
Guo, Cong
Zhang, Junyao
Shan, Haoxuan
Xu, Yifan
Zhang, Ziyue
Liu, Yudong
Wang, Qinsi
Zhou, Changchun
Li, Hai "Helen"
Chen, Yiran
Hardware Architecture
Vision-Language Models (VLMs) have demonstrated strong performance on tasks such as video captioning and visual question answering. However, their growing scale and video-level inputs lead to significant computational and memory overhead, posing challenges for real-time deployment on hardware accelerators. While prior work attempts to reduce redundancy via token pruning or merging, these methods typically operate at coarse granularity and incur high runtime overhead due to global token-level operations. In this study, we propose Focus, a Streaming Concentration Architecture that efficiently accelerates VLM inference through progressive, fine-grained redundancy elimination. Focus introduces a multilevel concentration paradigm that hierarchically compresses vision-language inputs at three levels: (1) semantic-guided token pruning based on textual prompts, (2) spatial-temporal block-level concentration using localized comparisons, and (3) vector-level redundancy removal via motion-aware matching. All concentration steps are tightly co-designed with the architecture to support streaming-friendly, on-chip execution. Focus leverages GEMM tiling, convolution-style layout, and cross-modal attention to minimize off-chip access while enabling high throughput. Implemented as a modular unit within a systolic-array accelerator, Focus achieves a 2.4x speedup and 3.3x reduction in energy, significantly outperforming state-of-the-art accelerators in both performance and energy efficiency. Full-stack implementation of Focus is open-sourced at https://github.com/dubcyfor3/Focus.
title Focus: A Streaming Concentration Architecture for Efficient Vision-Language Models
topic Hardware Architecture
url https://arxiv.org/abs/2512.14661