Topology-Aware Layer Pruning for Large Vision-Language Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zheng, Pengcheng, Zhang, Chaoning, Wen, Ya, Liu, Wang, Sun, Qigan, Mo, Jiarong, Zhang, Jiaquan, Lee, Jewon, Kim, Tae-Ho, Liu, Kuien, Li, Tianyu, Qin, Caiyan, Yang, Yang
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910142327422976
author Zheng, Pengcheng
Zhang, Chaoning
Wen, Ya
Liu, Wang
Sun, Qigan
Mo, Jiarong
Zhang, Jiaquan
Lee, Jewon
Kim, Tae-Ho
Liu, Kuien
Li, Tianyu
Qin, Caiyan
Yang, Yang
author_facet Zheng, Pengcheng
Zhang, Chaoning
Wen, Ya
Liu, Wang
Sun, Qigan
Mo, Jiarong
Zhang, Jiaquan
Lee, Jewon
Kim, Tae-Ho
Liu, Kuien
Li, Tianyu
Qin, Caiyan
Yang, Yang
contents Large Language Models (LLMs) have demonstrated strong capabilities in natural language understanding and reasoning, while recent extensions that incorporate visual inputs enable them to process multimodal information. Despite these advances, Large Vision-Language Models (LVLMs) incur substantial computational and memory costs, hindering deployment in resource-constrained scenarios. Existing layer pruning methods typically rely on local similarity metrics or static proxy signals, failing to capture the global and dynamic evolution of representations across model depth, which often leads to the removal of transition-critical layers. To address this limitation, we propose a topology-aware layer pruning framework for LVLMs. Specifically, we represent layer wise hidden states as point clouds and models their evolution using \textit{simplicial complexes}. By leveraging \textit{zigzag persistent homology}, we quantify inter-layer topological consistency and enable adaptive pruning that preserves critical representational transitions. Extensive experiments on diverse multimodal benchmarks demonstrate that the proposed framework consistently outperforms existing pruning methods across a wide range of sparsity ratios. Our code is available at https://github.com/zpc456/TopoVLM.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16502
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Topology-Aware Layer Pruning for Large Vision-Language Models
Zheng, Pengcheng
Zhang, Chaoning
Wen, Ya
Liu, Wang
Sun, Qigan
Mo, Jiarong
Zhang, Jiaquan
Lee, Jewon
Kim, Tae-Ho
Liu, Kuien
Li, Tianyu
Qin, Caiyan
Yang, Yang
Computer Vision and Pattern Recognition
Large Language Models (LLMs) have demonstrated strong capabilities in natural language understanding and reasoning, while recent extensions that incorporate visual inputs enable them to process multimodal information. Despite these advances, Large Vision-Language Models (LVLMs) incur substantial computational and memory costs, hindering deployment in resource-constrained scenarios. Existing layer pruning methods typically rely on local similarity metrics or static proxy signals, failing to capture the global and dynamic evolution of representations across model depth, which often leads to the removal of transition-critical layers. To address this limitation, we propose a topology-aware layer pruning framework for LVLMs. Specifically, we represent layer wise hidden states as point clouds and models their evolution using \textit{simplicial complexes}. By leveraging \textit{zigzag persistent homology}, we quantify inter-layer topological consistency and enable adaptive pruning that preserves critical representational transitions. Extensive experiments on diverse multimodal benchmarks demonstrate that the proposed framework consistently outperforms existing pruning methods across a wide range of sparsity ratios. Our code is available at https://github.com/zpc456/TopoVLM.
title Topology-Aware Layer Pruning for Large Vision-Language Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.16502