Training-free Token Reduction for Vision Mamba

• URL: http://arxiv.org/abs/2507.14042v1
• Date: Fri, 18 Jul 2025 16:11:28 GMT
• Title: Training-free Token Reduction for Vision Mamba
• Authors: Qiankun Ma, Ziyao Zhang, Chi Su, Jie Chen, Zhen Song, Hairong Zheng, Wen Gao,
• Abstract summary: Vision Mamba has emerged as a strong competitor to Vision Transformers (ViTs)<n>Applying token reduction techniques for ViTs to Vision Mamba leads to significant performance degradation.<n>We propose MTR, a training-free textbfMamba textbfToken textbfReduction framework.
• Score: 21.451182941570394
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision Mamba has emerged as a strong competitor to Vision Transformers (ViTs) due to its ability to efficiently capture long-range dependencies with linear computational complexity. While token reduction, an effective compression technique in ViTs, has rarely been explored in Vision Mamba. Exploring Vision Mamba's efficiency is essential for enabling broader applications. However, we find that directly applying existing token reduction techniques for ViTs to Vision Mamba leads to significant performance degradation. This is primarily because Mamba is a sequence model without attention mechanisms, whereas most token reduction techniques for ViTs rely on attention mechanisms for importance measurement and overlook the order of compressed tokens. In this paper, we investigate a Mamba structure-aware importance score to evaluate token importance in a simple and effective manner. Building on this score, we further propose MTR, a training-free \textbf{M}amba \textbf{T}oken \textbf{R}eduction framework. Without the need for training or additional tuning parameters, our method can be seamlessly integrated as a plug-and-play component across various Mamba models. Extensive experiments demonstrate that our approach significantly reduces computational workload while minimizing performance impact across various tasks and multiple backbones. Notably, MTR reduces FLOPs by approximately 40\% on the Vim-B backbone, with only a 1.6\% drop in ImageNet performance without retraining.

Related papers

• Dynamic Vision Mamba [41.84910346271891]
  Mamba-based vision models have gained extensive attention as a result of being computationally more efficient than attention-based models.<n>For token redundancy, we analytically find that early token pruning methods will result in inconsistency between training and inference.<n>For block redundancy, we allow each image to select SSM blocks dynamically based on an empirical observation that the inference speed of Mamba-based vision models is largely affected by the number of SSM blocks.
  arXiv  Detail & Related papers  (2025-04-07T07:31:28Z)
• Faster Vision Mamba is Rebuilt in Minutes via Merged Token Re-training [25.165300765309798]
  Empirically, pruned Vims only drop up to 0.9% accuracy on ImageNet-1K, recovered by our proposed framework R-MeeTo.<n> Vim-Ti/S/B are re-trained within 5/7/17 minutes, and Vim-S only drops 1.3% with 1.2x (up to 1.5x) speed up in inference.
  arXiv  Detail & Related papers  (2024-12-17T02:56:35Z)
• TinyViM: Frequency Decoupling for Tiny Hybrid Vision Mamba [11.176993272867396]
  Mamba has shown great potential for computer vision due to its linear complexity. Existing lightweight Mamba-based backbones cannot demonstrate performance that matches Convolution or Transformer-based methods. By integrating mobile-friendly convolution and efficient Laplace mixer, we build a series of tiny hybrid vision Mamba called TinyViM.
  arXiv  Detail & Related papers  (2024-11-26T14:34:36Z)
• MobileMamba: Lightweight Multi-Receptive Visual Mamba Network [51.33486891724516]
  Previous research on lightweight models has primarily focused on CNNs and Transformer-based designs. We propose the MobileMamba framework, which balances efficiency and performance. MobileMamba achieves up to 83.6% on Top-1, surpassing existing state-of-the-art methods.
  arXiv  Detail & Related papers  (2024-11-24T18:01:05Z)
• MAP: Unleashing Hybrid Mamba-Transformer Vision Backbone's Potential with Masked Autoregressive Pretraining [23.37555991996508]
  We propose Masked Autoregressive Pretraining (MAP) to pretrain a hybrid Mamba-Transformer vision backbone network.<n> Experimental results show that the hybrid Mamba-Transformer vision backbone network pretrained with MAP significantly outperforms other pretraining strategies.
  arXiv  Detail & Related papers  (2024-10-01T17:05:08Z)
• Famba-V: Fast Vision Mamba with Cross-Layer Token Fusion [10.854742185190482]
  Mamba and Vision Mamba (Vim) models have shown their potential as an alternative to methods based on Transformer architecture. This work introduces Fast Mamba for Vision (Famba-V), a cross-layer token fusion technique to enhance the training efficiency of Vim models.
  arXiv  Detail & Related papers  (2024-09-15T18:02:26Z)
• CAS-ViT: Convolutional Additive Self-attention Vision Transformers for Efficient Mobile Applications [73.80247057590519]
  Vision Transformers (ViTs) mark a revolutionary advance in neural networks with their token mixer's powerful global context capability.<n>We introduce CAS-ViT: Convolutional Additive Self-attention Vision Transformers, to achieve a balance between efficiency and performance in mobile applications.<n>Our model achieves 83.0%/84.1% top-1 with only 12M/21M parameters on ImageNet-1K.
  arXiv  Detail & Related papers  (2024-08-07T11:33:46Z)
• MambaVision: A Hybrid Mamba-Transformer Vision Backbone [54.965143338206644]
  We propose a novel hybrid Mamba-Transformer backbone, MambaVision, specifically tailored for vision applications.<n>We show that equipping the Mamba architecture with self-attention blocks in the final layers greatly improves its capacity to capture long-range spatial dependencies.<n>For classification on the ImageNet-1K dataset, MambaVision variants achieve state-of-the-art (SOTA) performance in terms of both Top-1 accuracy and throughput.
  arXiv  Detail & Related papers  (2024-07-10T23:02:45Z)
• Demystify Mamba in Vision: A Linear Attention Perspective [72.93213667713493]
  Mamba is an effective state space model with linear computation complexity.<n>We show that Mamba shares surprising similarities with linear attention Transformer.<n>We propose a Mamba-Inspired Linear Attention (MILA) model by incorporating the merits of these two key designs into linear attention.
  arXiv  Detail & Related papers  (2024-05-26T15:31:09Z)
• Swin-UMamba: Mamba-based UNet with ImageNet-based pretraining [85.08169822181685]
  This paper introduces a novel Mamba-based model, Swin-UMamba, designed specifically for medical image segmentation tasks. Swin-UMamba demonstrates superior performance with a large margin compared to CNNs, ViTs, and latest Mamba-based models.
  arXiv  Detail & Related papers  (2024-02-05T18:58:11Z)
• VMamba: Visual State Space Model [98.0517369083152]
  We adapt Mamba, a state-space language model, into VMamba, a vision backbone with linear time complexity.<n>At the core of VMamba is a stack of Visual State-Space (VSS) blocks with the 2D Selective Scan (SS2D) module.
  arXiv  Detail & Related papers  (2024-01-18T17:55:39Z)
• Self-slimmed Vision Transformer [52.67243496139175]
  Vision transformers (ViTs) have become the popular structures and outperformed convolutional neural networks (CNNs) on various vision tasks. We propose a generic self-slimmed learning approach for vanilla ViTs, namely SiT. Specifically, we first design a novel Token Slimming Module (TSM), which can boost the inference efficiency of ViTs.
  arXiv  Detail & Related papers  (2021-11-24T16:48:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.