GammaGL: A Multi-Backend library for Graph Neural Networks

What is it about?

GammaGL provides support for the implementation of Graph Neural Networks. Users can run the same code on the four deep-learning backends with a single line of code change. We also provide experiments to show the efficiency of our library.

Featured Image

Why is it important?

We support multiple deep-learning frameworks. Users may prefer a certain framework for its unique feature. However, switching different backends requires many lines of code modification. In GammaGL, users can enjoy any of the four frameworks with only a single line of code change. Our experiments also prove that our operators are efficient.