BriefGPT.xyz
Apr, 2018
神经网络的训练与推理的价值感知量化
Value-aware Quantization for Training and Inference of Neural Networks
HTML
PDF
Eunhyeok Park, Sungjoo Yoo, Peter Vajda
TL;DR
提出了一种新的价值感知量化方法，通过将大多数数据应用极低的精度并单独处理一小部分高精度数据以减少总量化误差。该方法可显著减少ResNet-152和Inception-v3的激活器内存成本，并能实现1％以下的top-1精度下降。
Abstract
We propose a novel
value-aware quantization
which applies aggressively
reduced precision
to the majority of data while separately handling a small amount of large data in high precision, which reduces total quant
→