AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Value-aware Quantization for Training and Inference of Neural Networks

Eunhyeok Park, Sungjoo Yoo, Peter Vajda

ECCV 2018 · Architectures and efficiency › Quantization · approach: Mixed-precision quantization · 200 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 2% large activations kept at high precision to 2-bit V-Quant (ablation)V-Quant ResNet-50 · ImageNet val (training with quantized activations)top-1 accuracy: 65.754 -> 75.638Improved
add 1% large activations to 1-bit V-Quant (ablation)V-Quant ResNet-50 · ImageNet val (training with quantized activations)top-1 accuracy: 5.302 -> 74.510Improved
add 2% large activations to 3-bit V-Quant (ablation)V-Quant ResNet-50 · ImageNet val (training with quantized activations)top-1 accuracy: 75.486 -> 75.920Improved
add 2% large activations to 3-bit RV-Quant (ablation)RV-Quant ResNet-50 · ImageNet val (training with quantized activations)top-1 accuracy: 75.156 -> 75.932Improved
add 1% large activations to 2-bit RV-Quant (ablation)RV-Quant ResNet-50 · ImageNet val (training with quantized activations)top-1 accuracy: 35.518 -> 75.338Improved
use RV-Quant mask sharing instead of V-Quant with separate bit mask (ablation)3-bit 2% ResNet-50 · ImageNet val (training with quantized activations)memory reduction of stored activations: 6.1X -> 7.5XImproved
start training in low precision (2,0) instead of full precision (ablation)(F)-(3,2)-(2,0) schedule · ImageNet val (activation annealing, ResNet-50)top-1 accuracy: 75.454 -> 50.360Hurt
anneal large activation ratio from 2% to 1% to 0% at 3 bits (ablation)(3,2)-(3,2)-(3,2) · ImageNet val (activation annealing, ResNet-50)top-1 accuracy: 75.720 vs similar to constant (3,2)No clear effect
anneal further to 2-bit at the end (ablation)(3,2)-(3,1)-(2,0) · ImageNet val (activation annealing, ResNet-50)top-1 accuracy: 0.584% smaller accuracyHurt
replace checkpointing with RV-Quant (3,2)ResNet-152 · ImageNet training activation memorymemory: 5.29GB -> 3.09GBImproved
replace checkpointing with RV-Quant (3,2)Inception-v3 · ImageNet training activation memorymemory: 3.87GB -> 1.79GBImproved
replace checkpointing with RV-QuantResNet-50 training runtime · ResNet-50 batch 64 on M40runtime overhead: 32.4% -> 8.8%Improved

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