r/MachineLearning - [D] Blog on mobile-optimized neural networks

I'll be covering light-weight model architectures, model compression techniques, and other optimizations in ML libraries; mostly from the perspective of targetting efficient deployment of DNNs.