AI Paper Atlas

Research map / Architectures and efficiency / Quantization

HAWQ: Hessian AWare Quantization of Neural Networks With Mixed-Precision

Zhen Dong, Zhewei Yao, Amir Gholami, Michael W. Mahoney, Kurt Keutzer

ICCV 2019 · Architectures and efficiency › Quantization · approach: Mixed-precision quantization · 564 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Hessian-aware mixed precision (vs direct mixed precision) (ablation)ResNet20 W2MP A4 · CIFAR-10accuracy (%): 90.34 -> 92.22Improved
add Hessian-aware mixed precision (vs direct mixed precision) (ablation)Inception-V3 W2MP A4MP · ImageNettop-1 (%): 69.76 -> 75.52Improved
add Hessian-aware mixed precision (vs direct mixed precision) (ablation)SqueezeNext W3MP A8 · ImageNettop-1 (%): 65.39 -> 68.02Improved
use Hessian-based precision order instead of reversed order (ablation)Inception-V3 · ImageNettop-1 (%): 66.72 -> 74.26Improved
use Hessian-based precision order instead of reversed order (ablation)Inception-V3 · ImageNetepochs to converge: 50 -> 30Improved
use descending Omega block-wise fine-tuning order instead of reverse order (ablation)Inception-V3 · ImageNetepochs to converge for a block: 50 -> 25Improved
use HAWQ instead of LQ-Nets 3-bitResNet20 · CIFAR-10accuracy (%): 91.60 -> 92.22Improved
use HAWQ instead of HAQ on ResNet50ResNet50 · ImageNettop-1 (%): 75.30 -> 75.48Improved
use HAWQ instead of RVQuantInception-V3 · ImageNettop-1 (%): 74.14 -> 75.52Improved
use HAWQ instead of DNASResNet20 · CIFAR-10activation compression: 1.00x -> 8.00xImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy