AdaTok: Adaptive Token Compression with Object-Aware Representations for Efficient Multimodal LLMs

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Zhang, Xinliang, Zhu, Lei, He, Hangzhou, Zeng, Shuang, Fu, Ourui, Hu, Jiakui, Yao, Zhengjian, Lu, Yanye
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918216016592896
author Zhang, Xinliang
Zhu, Lei
He, Hangzhou
Zeng, Shuang
Fu, Ourui
Hu, Jiakui
Yao, Zhengjian
Lu, Yanye
author_facet Zhang, Xinliang
Zhu, Lei
He, Hangzhou
Zeng, Shuang
Fu, Ourui
Hu, Jiakui
Yao, Zhengjian
Lu, Yanye
contents Multimodal Large Language Models (MLLMs) have demonstrated substantial value in unified text-image understanding and reasoning, primarily by converting images into sequences of patch-level tokens that align with their architectural paradigm. However, patch-level tokenization leads to a quadratic growth in image tokens, burdening MLLMs' understanding and reasoning with enormous computation and memory. Additionally, the traditional patch-wise scanning tokenization workflow misaligns with the human vision cognition system, further leading to hallucination and computational redundancy. To address this issue, we propose an object-level token merging strategy for Adaptive Token compression, revealing the consistency with human vision system. The experiments are conducted on multiple comprehensive benchmarks, which show that our approach averagely, utilizes only 10% tokens while achieving almost 96% of the vanilla model's performance. More extensive experimental results in comparison with relevant works demonstrate the superiority of our method in balancing compression ratio and performance. Our code will be available.
format Preprint
id arxiv_https___arxiv_org_abs_2511_14169
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AdaTok: Adaptive Token Compression with Object-Aware Representations for Efficient Multimodal LLMs
Zhang, Xinliang
Zhu, Lei
He, Hangzhou
Zeng, Shuang
Fu, Ourui
Hu, Jiakui
Yao, Zhengjian
Lu, Yanye
Computer Vision and Pattern Recognition
Artificial Intelligence
Multimodal Large Language Models (MLLMs) have demonstrated substantial value in unified text-image understanding and reasoning, primarily by converting images into sequences of patch-level tokens that align with their architectural paradigm. However, patch-level tokenization leads to a quadratic growth in image tokens, burdening MLLMs' understanding and reasoning with enormous computation and memory. Additionally, the traditional patch-wise scanning tokenization workflow misaligns with the human vision cognition system, further leading to hallucination and computational redundancy. To address this issue, we propose an object-level token merging strategy for Adaptive Token compression, revealing the consistency with human vision system. The experiments are conducted on multiple comprehensive benchmarks, which show that our approach averagely, utilizes only 10% tokens while achieving almost 96% of the vanilla model's performance. More extensive experimental results in comparison with relevant works demonstrate the superiority of our method in balancing compression ratio and performance. Our code will be available.
title AdaTok: Adaptive Token Compression with Object-Aware Representations for Efficient Multimodal LLMs
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2511.14169