This is a gentle introduction to building a Graph Neural Network with PyTorch Geometric, the popular library for geometric deep learning. The tutorial will also cover some streamlined interpretability tools for analyzing the decision process of a GNN.
The tutorial is a Python notebook that you can
To run the notebook locally, you need the following libraries:
- PyTorch
- PyTorch Geometric
- Pytorch Lightning
- Captum
- Networkx
This tutorial is adapted from the tutorial of Simone Scardapane (slides, notebook).