Pack-PTQ: Advancing Post-training Quantization of Neural Networks by Pack-wise Reconstruction

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Changjun, Jiang, Runqing, Song, Zhuo, Yu, Pengpeng, Zhang, Ye, Guo, Yulan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915268247158784
author Li, Changjun
Jiang, Runqing
Song, Zhuo
Yu, Pengpeng
Zhang, Ye
Guo, Yulan
author_facet Li, Changjun
Jiang, Runqing
Song, Zhuo
Yu, Pengpeng
Zhang, Ye
Guo, Yulan
contents Post-training quantization (PTQ) has evolved as a prominent solution for compressing complex models, which advocates a small calibration dataset and avoids end-to-end retraining. However, most existing PTQ methods employ block-wise reconstruction, which neglects cross-block dependency and exhibits a notable accuracy drop in low-bit cases. To address these limitations, this paper presents a novel PTQ method, dubbed Pack-PTQ. First, we design a Hessian-guided adaptive packing mechanism to partition blocks into non-overlapping packs, which serve as the base unit for reconstruction, thereby preserving the cross-block dependency and enabling accurate quantization parameters estimation. Second, based on the pack configuration, we propose a mixed-precision quantization approach to assign varied bit-widths to packs according to their distinct sensitivities, thereby further enhancing performance. Extensive experiments on 2D image and 3D point cloud classification tasks, using various network architectures, demonstrate the superiority of our method over the state-of-the-art PTQ methods.
format Preprint
id arxiv_https___arxiv_org_abs_2505_00259
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Pack-PTQ: Advancing Post-training Quantization of Neural Networks by Pack-wise Reconstruction
Li, Changjun
Jiang, Runqing
Song, Zhuo
Yu, Pengpeng
Zhang, Ye
Guo, Yulan
Computer Vision and Pattern Recognition
Artificial Intelligence
Post-training quantization (PTQ) has evolved as a prominent solution for compressing complex models, which advocates a small calibration dataset and avoids end-to-end retraining. However, most existing PTQ methods employ block-wise reconstruction, which neglects cross-block dependency and exhibits a notable accuracy drop in low-bit cases. To address these limitations, this paper presents a novel PTQ method, dubbed Pack-PTQ. First, we design a Hessian-guided adaptive packing mechanism to partition blocks into non-overlapping packs, which serve as the base unit for reconstruction, thereby preserving the cross-block dependency and enabling accurate quantization parameters estimation. Second, based on the pack configuration, we propose a mixed-precision quantization approach to assign varied bit-widths to packs according to their distinct sensitivities, thereby further enhancing performance. Extensive experiments on 2D image and 3D point cloud classification tasks, using various network architectures, demonstrate the superiority of our method over the state-of-the-art PTQ methods.
title Pack-PTQ: Advancing Post-training Quantization of Neural Networks by Pack-wise Reconstruction
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2505.00259