ERQ: Error Reduction for Post-Training Quantization of Vision Transformers

Yunshan Zhong, Jiawei Hu, You Huang, Yuxin Zhang, Rongrong Ji
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:61664-61680, 2024.

Abstract

Post-training quantization (PTQ) for vision transformers (ViTs) has garnered significant attention due to its efficiency in compressing models. However, existing methods typically overlook the intricate interdependence between quantized weight and activation, leading to considerable quantization error. In this paper, we propose ERQ, a two-step PTQ approach meticulously crafted to sequentially reduce the quantization error arising from activation and weight quantization. ERQ first introduces Activation quantization error reduction (Aqer) that strategically formulates the minimization of activation quantization error as a Ridge Regression problem, tackling it by updating weights with full-precision. Subsequently, ERQ introduces Weight quantization error reduction (Wqer) that adopts an iterative approach to mitigate the quantization error induced by weight quantization. In each iteration, an empirically derived, efficient proxy is employed to refine the rounding directions of quantized weights, coupled with a Ridge Regression solver to curtail weight quantization error. Experimental results attest to the effectiveness of our approach. Notably, ERQ surpasses the state-of-the-art GPTQ by 22.36% in accuracy for W3A4 ViT-S.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-zhong24a, title = {{ERQ}: Error Reduction for Post-Training Quantization of Vision Transformers}, author = {Zhong, Yunshan and Hu, Jiawei and Huang, You and Zhang, Yuxin and Ji, Rongrong}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {61664--61680}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/zhong24a/zhong24a.pdf}, url = {https://proceedings.mlr.press/v235/zhong24a.html}, abstract = {Post-training quantization (PTQ) for vision transformers (ViTs) has garnered significant attention due to its efficiency in compressing models. However, existing methods typically overlook the intricate interdependence between quantized weight and activation, leading to considerable quantization error. In this paper, we propose ERQ, a two-step PTQ approach meticulously crafted to sequentially reduce the quantization error arising from activation and weight quantization. ERQ first introduces Activation quantization error reduction (Aqer) that strategically formulates the minimization of activation quantization error as a Ridge Regression problem, tackling it by updating weights with full-precision. Subsequently, ERQ introduces Weight quantization error reduction (Wqer) that adopts an iterative approach to mitigate the quantization error induced by weight quantization. In each iteration, an empirically derived, efficient proxy is employed to refine the rounding directions of quantized weights, coupled with a Ridge Regression solver to curtail weight quantization error. Experimental results attest to the effectiveness of our approach. Notably, ERQ surpasses the state-of-the-art GPTQ by 22.36% in accuracy for W3A4 ViT-S.} }
Endnote
%0 Conference Paper %T ERQ: Error Reduction for Post-Training Quantization of Vision Transformers %A Yunshan Zhong %A Jiawei Hu %A You Huang %A Yuxin Zhang %A Rongrong Ji %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-zhong24a %I PMLR %P 61664--61680 %U https://proceedings.mlr.press/v235/zhong24a.html %V 235 %X Post-training quantization (PTQ) for vision transformers (ViTs) has garnered significant attention due to its efficiency in compressing models. However, existing methods typically overlook the intricate interdependence between quantized weight and activation, leading to considerable quantization error. In this paper, we propose ERQ, a two-step PTQ approach meticulously crafted to sequentially reduce the quantization error arising from activation and weight quantization. ERQ first introduces Activation quantization error reduction (Aqer) that strategically formulates the minimization of activation quantization error as a Ridge Regression problem, tackling it by updating weights with full-precision. Subsequently, ERQ introduces Weight quantization error reduction (Wqer) that adopts an iterative approach to mitigate the quantization error induced by weight quantization. In each iteration, an empirically derived, efficient proxy is employed to refine the rounding directions of quantized weights, coupled with a Ridge Regression solver to curtail weight quantization error. Experimental results attest to the effectiveness of our approach. Notably, ERQ surpasses the state-of-the-art GPTQ by 22.36% in accuracy for W3A4 ViT-S.
APA
Zhong, Y., Hu, J., Huang, Y., Zhang, Y. & Ji, R.. (2024). ERQ: Error Reduction for Post-Training Quantization of Vision Transformers. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:61664-61680 Available from https://proceedings.mlr.press/v235/zhong24a.html.

Related Material