A typical machine learning development cycle maximizes performance during model training and then minimizes the memory and area footprint of the trained model for deployment on processing cores, ...