Weight Group-wise Post-Training Quantization for Medical Foundation Model
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917394423742464 |
|---|---|
| author | Chen, Yineng Huang, Peng Zhang, Aozhong Guo, Hui Yin, Penghang Hu, Shu Lin, Shao Li, Xin Kao, Tzu-Jen Prabhakaran, Balakrishnan Chang, MingChing Wang, Xin |
| author_facet | Chen, Yineng Huang, Peng Zhang, Aozhong Guo, Hui Yin, Penghang Hu, Shu Lin, Shao Li, Xin Kao, Tzu-Jen Prabhakaran, Balakrishnan Chang, MingChing Wang, Xin |
| contents | Foundation models have achieved remarkable results in medical image analysis. However, its large network architecture and high computational complexity significantly impact inference speed, limiting its application on terminal medical devices. Quantization, a technique that compresses models into low-bit versions, is a solution to this challenge. In this paper, we propose a post-training quantization algorithm, Permutation-COMQ. It eliminates the need for backpropagation by using simple dot products and rounding operations, thereby removing hyperparameter tuning and simplifying the process. Additionally, we introduce a weight-aware strategy that reorders the weight within each layer to address the accuracy degradation induced by channel-wise scaling during quantization, while preserving channel structure. Experiments demonstrate that our method achieves the best results in 2-bit, 4-bit, and 8-bit quantization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_07674 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Weight Group-wise Post-Training Quantization for Medical Foundation Model Chen, Yineng Huang, Peng Zhang, Aozhong Guo, Hui Yin, Penghang Hu, Shu Lin, Shao Li, Xin Kao, Tzu-Jen Prabhakaran, Balakrishnan Chang, MingChing Wang, Xin Computer Vision and Pattern Recognition Foundation models have achieved remarkable results in medical image analysis. However, its large network architecture and high computational complexity significantly impact inference speed, limiting its application on terminal medical devices. Quantization, a technique that compresses models into low-bit versions, is a solution to this challenge. In this paper, we propose a post-training quantization algorithm, Permutation-COMQ. It eliminates the need for backpropagation by using simple dot products and rounding operations, thereby removing hyperparameter tuning and simplifying the process. Additionally, we introduce a weight-aware strategy that reorders the weight within each layer to address the accuracy degradation induced by channel-wise scaling during quantization, while preserving channel structure. Experiments demonstrate that our method achieves the best results in 2-bit, 4-bit, and 8-bit quantization. |
| title | Weight Group-wise Post-Training Quantization for Medical Foundation Model |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.07674 |