I&S-ViT: An Inclusive & Stable Method for Post-Training ViTs Quantization
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-02 · 48 (2) · 1063-1080
DOI: 10.1109/tpami.2025.3610466 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-02. Verfasst von Yunshan Zhong und 4 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 87 Literaturangaben. Gefördert von: National Key R&D Program of China, National Science Fund for Distinguished Young Scholars, National Natural Science Foundation of China.
Abstract (Originalsprache)
Albeit the scalable performance of vision transformers (ViTs), the dense computational costs undermine their position in industrial applications. Post-training quantization (PTQ), tuning ViTs with a tiny dataset and running in a low-bit format, well addresses the cost issue but unluckily bears more performance drops in lower-bit cases. In this paper, we introduce I&S-ViT, a novel method that regulates the PTQ of ViTs in an inclusive and stable fashion. I&S-ViT first identifies two issues in the PTQ of ViTs: (1) Quantization inefficiency in the prevalent log2 quantizer for post-Softmax activations; (2) Rugged and magnified loss landscape in coarse-grained quantization granularity for post-LayerNorm activations. Then, I&S-ViT addresses these issues by introducing: (1) A novel shift-uniform-log2 quantizer (SULQ) that incorporates a shift mechanism followed by uniform quantization to achieve both an inclusive domain representation and accurate distribution approximation; (2) A three-stage smooth optimization strategy (SOS) that amalgamates the strengths of channel-wise and layer-wise quantization to enable stable learning. Comprehensive evaluations across diverse vision tasks validate I&S-ViT's superiority over existing PTQ of ViTs methods, particularly in low-bit scenarios. For instance, I&S-ViT elevates the performance of W3A3 ViT-B by an impressive 50.68%.
Förderung
- National Key R&D Program of China
- National Science Fund for Distinguished Young Scholars
- National Natural Science Foundation of China
- Natural Science Foundation of Fujian Province of China
Zitieren
APA
Zhong, Y., Hu, J., Lin, M., Chen, M., & Ji, R. (2026). I&S-ViT: An Inclusive & Stable Method for Post-Training ViTs Quantization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(2), 1063-1080. https://doi.org/10.1109/tpami.2025.3610466
Vancouver
Zhong Y, Hu J, Lin M, Chen M, Ji R. I&S-ViT: An Inclusive & Stable Method for Post-Training ViTs Quantization. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(2):1063-1080. doi:10.1109/tpami.2025.3610466
BibTeX
@article{Zhong2026,
author = {Zhong, Yunshan and Hu, Jiawei and Lin, Mingbao and Chen, Mengzhao and Ji, Rongrong},
title = {I&S-ViT: An Inclusive & Stable Method for Post-Training ViTs Quantization},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2026},
volume = {48},
number = {2},
pages = {1063--1080},
doi = {10.1109/tpami.2025.3610466}
}
EUVIMED