BriefGPT.xyz
Dec, 2022
QFT：通过快速联合微调所有自由度进行后训练量化
QFT: Post-training quantization via fast joint finetuning of all degrees of freedom
HTML
PDF
Alex Finkelstein, Ella Fuchs, Idan Tal, Mark Grobman, Niv Vosco...
TL;DR
提出了一种硬件意识的量化网络参数化方法——量化感知微调（QFT），可以通过联合端对端微调实现一步到位的量化，获得与最优结果相当的4位权重量化结果。
Abstract
The
post-training quantization
(PTQ) challenge of bringing quantized
neural net accuracy
close to original has drawn much attention driven by industry demand. Many of the methods emphasize optimization of a speci
→