Pith. sign in

REVIEW 2 cited by

Faster Vision Mamba is Rebuilt in Minutes via Merged Token Re-training

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2412.12496 v4 pith:2FMSNSFE submitted 2024-12-17 cs.CV cs.AI

Mingjia Shi , Yuhao Zhou , Ruiji Yu , Zekai Li , Zhiyuan Liang , Xuanlei Zhao , Xiaojiang Peng , Shanmukha Ramakrishna Vedantam
show 3 more authors
Wangbo Zhao Kai Wang Yang You
This is my paper
classification cs.CVcs.AI
keywords mambatokenvisionaccuracycompressionmergingminutesonly
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Vision Mamba has shown close to state of the art performance on computer vision tasks, drawing much interest in increasing it's efficiency. A promising approach is token reduction (that has been successfully implemented in ViTs). Pruning informative tokens in Mamba leads to a high loss of key knowledge and degraded performance. An alternative, of merging tokens preserves more information than pruning, also suffers for large compression ratios. Our key insight is that a quick round of retraining after token merging yeilds robust results across various compression ratios. Empirically, pruned Vims only drop up to 0.9% accuracy on ImageNet-1K, recovered by our proposed framework R-MeeTo in our main evaluation. We show how simple and effective the fast recovery can be achieved at minute-level, in particular, a 35.9% accuracy spike over 3 epochs of training on Vim-Ti. Moreover, 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.

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. QuarterMap: Efficient Post-Training Token Pruning for Visual State Space Models

    cs.CV 2025-07 conditional novelty 5.0 of 10

    QuarterMap prunes spatial activations before VMamba's four-directional scan and upsamples after, yielding up to 1.11x throughput with under 1% accuracy loss on ImageNet classification.

  2. Exploring Kolmogorov-Arnold Network Expansions in Vision Transformers for Mitigating Catastrophic Forgetting in Continual Learning

    cs.CV 2025-07 reject novelty 3.0 of 10

    KAN-based ViTs show slight average incremental accuracy gains over MLP-ViTs in continual learning, but the paper's own data show worse forgetting on CIFAR-100 and worse last-task accuracy on MNIST.

Pith tools