AI Paper Atlas

Research map / Image generation / GANs

Generative Low-bitwidth Data Free Quantization

Shoukai Xu, Haokun Li, Bohan Zhuang, Jing Liu, Jiezhang Cao, Chuangrun Liang, Mingkui Tan

ECCV 2020 · Image generation › GANs · approach: GAN latent optimization and evaluation · 143 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 CE loss to generator training (without BNS loss) (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 30.70 -> 54.51Improved
add BNS loss to generator training (without CE loss) (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 30.70 -> 44.40Improved
add BNS loss on top of CE loss for generator (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 54.51 -> 63.91Improved
add CE loss on top of BNS loss for generator (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 44.40 -> 63.91Improved
add KD loss on top of CE loss for quantized model (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 55.55 -> 63.91Improved
add CE loss on top of KD loss for quantized model (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 62.98 -> 63.91Improved
add fixed batch normalization statistics (FBNS) (ablation)GDFQ · CIFAR-10 ResNet-20 W4A4top-1 accuracy: 89.21 -> 90.23Improved
add fixed batch normalization statistics (FBNS) (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 61.12 -> 63.91Improved
alternate G and Q training instead of separate training (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 61.81 -> 63.91Improved
never stop generator training instead of stopping at threshold 99.50% (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 59.67 -> 63.91Improved
raise generator stopping threshold from 90.00% to 99.50% (ablation)GDFQ · CIFAR-100 ResNet-20 W4A4top-1 accuracy: 57.20 -> 59.67Improved
use fake data GDFQ instead of ZeroQGDFQ · ImageNet ResNet-18 W4A4top-1 accuracy: 26.04 -> 60.60Improved

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