Architectural responsibility “So, as a user of deep learning you really are piecing together layers, thinking about the data you will train them on… and then doing the training.