Saved in:
Bibliographic Details
Main Authors: Zhang, Xuan, Du, Cunxiao, Yu, Sicheng, Wu, Jiawei, Zhang, Fengzhuo, Gao, Wei, Liu, Qian
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.19155
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916021558837248
author Zhang, Xuan
Du, Cunxiao
Yu, Sicheng
Wu, Jiawei
Zhang, Fengzhuo
Gao, Wei
Liu, Qian
author_facet Zhang, Xuan
Du, Cunxiao
Yu, Sicheng
Wu, Jiawei
Zhang, Fengzhuo
Gao, Wei
Liu, Qian
contents Due to the auto-regressive nature of current video large language models (Video-LLMs), the inference latency increases as the input sequence length grows, posing challenges for the efficient processing of video sequences that are usually very long. We observe that during decoding, the attention scores of most tokens in Video-LLMs tend to be sparse and concentrated, with only certain tokens requiring comprehensive full attention. Based on this insight, we introduce Sparse-to-Dense (StD), a novel decoding strategy that integrates two distinct modules: one leveraging sparse top-K attention and the other employing dense full attention. These modules collaborate to accelerate Video-LLMs without loss. The fast (sparse) model speculatively decodes multiple tokens, while the slow (dense) model verifies them in parallel. StD is a tuning-free, plug-and-play solution that achieves up to a 1.94$\times$ walltime speedup in video processing. It maintains model performance while enabling a seamless transition from a standard Video-LLM to a sparse Video-LLM with minimal code modifications.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19155
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sparse-to-Dense: A Free Lunch for Lossless Acceleration of Video Understanding in LLMs
Zhang, Xuan
Du, Cunxiao
Yu, Sicheng
Wu, Jiawei
Zhang, Fengzhuo
Gao, Wei
Liu, Qian
Computer Vision and Pattern Recognition
Computation and Language
Due to the auto-regressive nature of current video large language models (Video-LLMs), the inference latency increases as the input sequence length grows, posing challenges for the efficient processing of video sequences that are usually very long. We observe that during decoding, the attention scores of most tokens in Video-LLMs tend to be sparse and concentrated, with only certain tokens requiring comprehensive full attention. Based on this insight, we introduce Sparse-to-Dense (StD), a novel decoding strategy that integrates two distinct modules: one leveraging sparse top-K attention and the other employing dense full attention. These modules collaborate to accelerate Video-LLMs without loss. The fast (sparse) model speculatively decodes multiple tokens, while the slow (dense) model verifies them in parallel. StD is a tuning-free, plug-and-play solution that achieves up to a 1.94$\times$ walltime speedup in video processing. It maintains model performance while enabling a seamless transition from a standard Video-LLM to a sparse Video-LLM with minimal code modifications.
title Sparse-to-Dense: A Free Lunch for Lossless Acceleration of Video Understanding in LLMs
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2505.19155