Train neural networks with joint quantization and pruning on both weights and activations using any …
mlzxy/qsparse