AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Rethinking Asymmetric Quantization: Hidden Symmetry in Vision Model Weights

Masafumi Mori, Shinya Gongyo, Mitsuru Ambai

CVPR 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of ViTs

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
replace zero-point AsymQ weights with dense SymQ + 98% sparse SymQ outliersRepQ-ViT · ImageNet ViT-B W3/A4top-1 accuracy (%): 26.98 -> 75.24Improved
replace AsymQ and DUQ with DASQERQ · ImageNet ViT-B W3/A4top-1 accuracy (%): 72.37 -> 78.44Improved
replace AsymQ with DASQRepQ-ViT · ImageNet ViT-B W4/A4top-1 accuracy (%): 68.48 -> 78.76Improved
replace AsymQ and DUQ with DASQERQ · ImageNet ViT-B W4/A4top-1 accuracy (%): 78.67 -> 80.04Improved
replace AsymQ with DASQQDrop · ImageNet ResNet-50 W2/A4top-1 accuracy (%): 70.08 -> 72.63Improved
replace AsymQ with DASQRepQ-ViT · COCO Mask R-CNN Swin-Tbox AP: 36.1 -> 41.5Improved
replace AsymQ with DASQRepQ-ViT · COCO Mask R-CNN Swin-Sbox AP: 44.2 -> 44.7Improved
replace AsymQ with DASQ on FPGAAsymQ W4/A4 · FPGA ViT-S ImageNettop-1 accuracy (%): up to +6.7%Improved
add 4-bit sparse matrix (98% sparsity) (ablation)RepQ-ViT SymQ vs AsymQ dense · ViT-S W4/A4top-1 accuracy (%): SymQ/AsymQ gap narrows to 0.19% at p=0.95Improved
use AsymQ instead of SymQ without sparse matrix (ablation)RepQ-ViT · ViT-S W5/A4top-1 accuracy (%): 71.77 -> 71.98No clear effect
use AsymQ instead of SymQ without sparse matrix (ablation)RepQ-ViT · ViT-S W4/A4top-1 accuracy (%): 61.56 -> 64.62Improved
increase sparse bit width from 4 to 8 (ablation)DASQ · FPGA ViT-Stop-1 accuracy (%): slight LUT/power increaseImproved

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