UML for deep learning architecture

Hello everyone,

can anybody share his experience with tools / methods for a clean documentation of Models?
I’ve tried using class diagram for my NTM.

Also I had the idea of drawing an activity diagram with a swimlane for each component to describe the flow of Tensors trough the model like this …

There’s a bunch of open source tools GitHub - szagoruyko/pytorchviz: A small package to create visualizations of PyTorch execution graphs

However if you have a better one in mind by all means create it, I think this is far from being a solved problem