The repository https://github.com/kochgroup/intro_pharma_ai contains different machine learning tutorials for pharmacists/chemists in german and english. The Jupyter notebooks can be run directly in Google Colab. For this SPP Tutorial, try the PyTorch Tutorial there, it has tasks to train the contents of the lecture, and a final exercise. In the first part, a NN is trained on the MNIST dataset (recognition of handdrawn digits), the final exercise instead uses data of the Tox21 challenge with fingerprints.
Note: In this notebook, we only use nn.Sequential. If you want, try it instead with a nn.Module class.
The notebook pytorch_geometric_introduction.ipynb in the folder with the same name goes through the installation of PyG, the loading of existing datasets, the creation of private datasets and the training of a simple GNN. This notebook does not contain any exercises.