Effective Quantization Methods for Recurrent Neural Networks

Qinyao He
Qinyao He
He Wen
He Wen
Yuheng Zou
Yuheng Zou

arXiv: Learning, Volume abs/1611.10176, 2016.

Cited by: 47|Bibtex|Views97
EI
Other Links: dblp.uni-trier.de|academic.microsoft.com|arxiv.org

Abstract:

Reducing bit-widths of weights, activations, and gradients of a Neural Network can shrink its storage size and memory usage, and also allow for faster training and inference by exploiting bitwise operations. However, previous attempts for quantization of RNNs show considerable performance degradation when using low bit-width weights and a...More

Code:

Data:

Your rating :
0

 

Tags
Comments