optimizing deep learning models with weight quantization c786ffc6d6c1