Why Scholars Are Diagramming Neural Network Models