EasyQuant: Post-training Quantization via Scale Optimization

Di Wu, Qi Tang, Yongle Zhao +3

The 8 bits quantization has been widely applied to accelerate network inference in various deep learning applications. There are two kinds of quantization methods, training-based q…