Fetching the paper…

Post-training 4-bit quantization of convolution networks for rapid-deployment · Around