Simple Symbolic Graphs in Python
However, these frameworks make several assumptions and optimisations in order to optimise for speed and memory. This often makes it harder to understand the inner workings of how these libraries work.
Sympyle is a simplified model library to demonstrate the working of computational graphs, and how backpropagation works on arbitrary 'networks'.
Examples and tutorials
All tutorials are under docs/source/tutorials and interactible at https://harveyslash.github.io/sympyle/ (under the tutorials section)