Can we add a example of a graph neural network for molecules ?
we can start there: https://uvadlc-notebooks.readthedocs.io/en/latest/tutorial_notebooks/JAX/tutorial7/GNN_overview.html or here: https://danielegrattarola.github.io/posts/2021-03-12/gnn-lecture-part-2.html
but ideally GIN, DMPNN and AttentiveFP are part of the best models so far: https://github.com/aimat-lab/gcnn_keras/tree/master/kgcnn/literature
I love this idea. We'd be happy to add one our example repo. Feel free to make a PR for this (or anyone who is interested)!
The idea seems cool, but this would require a lot of work to implement all message passing functions in C++, scattering etc. Are there any sparse operations implemented yet in MLX?