... and Pytorch Geometric
The key component of Graph Neural Networks are obviously node features. But how do you include edge information into the message passing mechanism? I talk about several possibilities in this video.
