AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Hardware-Centric AutoML for Mixed-Precision Quantization

Kuan Wang, Zhijian Liu, Yujun Lin, Ji Lin, Song Han

IJCV 2020 · Architectures and efficiency › Quantization · approach: Mixed-precision quantization · 18 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
use learned flexible per-layer bitwidths instead of uniform 4-bit PACT at same latencyMobileNet-V1 · BISMO edge latency-constrainedtop-1 accuracy (%): 62.44 -> 67.40Improved
use learned flexible bitwidths instead of uniform 4-bit PACT at same latencyMobileNet-V1 · BitFusion latency-constrainedtop-1 accuracy (%): 62.44 -> 67.45Improved
use learned flexible bitwidths instead of uniform 4-bit PACT at same energyMobileNet-V1 · BitFusion energy-constrainedtop-1 accuracy (%): 62.44 -> 64.78Improved
use learned flexible bitwidths instead of Deep Compression 2-bit at same model sizeMobileNet-V1 · ImageNet model size constrainttop-1 accuracy (%): 37.62 -> 57.14Improved
use flexible bitwidths instead of uniform 8-bit quantizationMobileNet-V1 · BISMO edge latencylatency: 96.20 ms -> 57.70 msImproved
use flexible bitwidths instead of uniform 8-bit quantizationMobileNet-V1 · BitFusion latencylatency: 20.08 -> 11.09 msImproved
use RL instead of evolutionary search for bitwidth policy (ablation)HAQ · BISMO edge 45 ms, MobileNet-V1top-1 accuracy (%): 65.73 -> 67.40Improved
use RL instead of Bayesian optimization for bitwidth policy (ablation)HAQ · BISMO edge 45 ms, MobileNet-V1top-1 accuracy (%): 66.28 -> 67.40; search time 74 hours -> 17 hoursImproved
transfer policy agent from MobileNet-V1 to V2 instead of searching directly on V2HAQ · BISMO edge 4-bit latencytop-1 accuracy (%): 66.99 -> 65.80Hurt
transfer V1-trained agent to V2 versus uniform 4-bit PACTHAQ · BISMO edge 4-bit latencytop-1 accuracy (%): 61.39 -> 65.80Improved
add direct hardware latency feedback instead of BitOPs proxy constraint (ablation)HAQ · BISMO edge, MobileNet-V2latency: 85.06 ms -> 66.92 msImproved
add hardware feedback instead of BitOPs constraint (tight budget) (ablation)HAQ · BISMO edge, MobileNet-V2top-1 accuracy (%): 70.29 -> 70.90Mixed

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