AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Leveraging Inter-Layer Dependency for Post -Training Quantization

Changbao Wang, DanDan Zheng, Yuanliu Liu, Liang Li

NeurIPS 2022 · Architectures and efficiency › Quantization · approach: Post-training weight quantization · 1 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 layer-wise activation regularization (AR) (ablation)NWQ · MobileNetV2 W2A2top-1 accuracy: 21.20 -> 26.42Improved
add layer-wise AR (ablation)NWQ · MnasNet W2A2top-1 accuracy: 30.75 -> 41.17Improved
add layer-wise AR (ablation)NWQ · ResNet18 W2A2top-1 accuracy: 57.94 -> 59.14Improved
coarser AR granularity (stage-wise vs layer-wise) (ablation)NWQ · MobileNetV2 W2A2top-1 accuracy: 26.42 -> 25.52Mixed
block-wise AR instead of layer-wise (ablation)NWQ · RegNet-3.2GF W2A2top-1 accuracy: 62.85 -> 61.31Hurt
add AutoAugment on top of AR (ablation)NWQ · MnasNet W2A2top-1 accuracy: 41.17 -> 42.92Improved
replace ASoftmax with RSeR (ablation)NWQ · MobileNetV2 W2A2top-1 accuracy: 26.42 -> 4.86Hurt
replace ASoftmax with RSeR (ablation)NWQ · ResNet18 W2A2top-1 accuracy: 59.14 -> 46.56Hurt
use proposed initialization instead of random (ablation)NWQ · ResNet18 W2A2top-1 accuracy: 55.33 -> 59.14Improved
final temperature tau_T from 1e-2 to 1e-1 (ablation)ASoftmax · MobileNetV2 W2A2top-1 accuracy: 26.42 -> 12.19Hurt
enlarge discrete range (n=-1, m=2) on small calibration set (ablation)ASoftmax · RegNet-3.2GF W2A2top-1 accuracy: 62.85 -> 61.68Hurt
scale up calibration set 10x (ablation)NWQ · MobileNetV2 W2A2top-1 accuracy: 26.42 -> 29.54Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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