BriefGPT.xyz
Feb, 2020
BitPruning: 学习位长进行激进而精确的量化
BitPruning: Learning Bitlengths for Aggressive and Accurate Quantization
HTML
PDF
Miloš Nikolić, Ghouthi Boukli Hacene, Ciaran Bannon, Alberto Delmas Lascorz, Matthieu Courbariaux...
TL;DR
通过提出一种惩罚体系惩罚大位长表示的正则化方法，我们可以在维持准确性的同时，在任意合适的层次上最小化推理位长。
Abstract
neural networks
have demonstrably achieved state-of-the art
accuracy
using low-
bitlength
integer
→