We are updating the README to be more descriptive; in the meantime, please see https://gm-neurips-2020.github.io/ or https://research.google/teams/graph-mining/

There are now more documents linked to in the README.md and an example you can try to run: https://github.com/google/graph-mining

It was hyped some years ago. There are plenty of legitimate applications of graphs, perhaps the library offers well optimized implementation of important algorithms. But the past hype around all things "graph" was not rational. As always, you can't solve all problems with a graph as you can't with a neural network or with any other structure/algorithm