A PyTorch implementation of "Incremental Network Quantization: Towards Lossless CNNs with Low-Precision Weights" - View it on GitHub