Explainer vs GeometricFlux

[bhalonen]

Hello,

I'm an experience Julia developer, and I am getting into GNN's. I am looking at your package and GeometricFlux, and I almost defaulted to GeometricFlux on popularity, but it appears you have many nice features that I might prefer.

Can you provide a "simple explanation" of why this package vs GeometricFlux (and vice versa)?

Great work by the way!

[AriMKatz]

#2

[CarloLucibello]

Well, if you have to perform graph-level tasks (e.g. graph classification), the choice is brainless,
GF.jl doesn't support batched graphs.

The performance comparison in #82 is another compelling argument inf favor of GNN.jl: on CPU, the Cora example runs 4x faster. On GPU, I couldn't manage to get it working for GF.jl, while GNN.jl gains another 10x.

Honestly, at the moment I cannot think of a single reason to recommend GF.jl over GNN.jl (performance, features, flexibility, simplicity, documentation, ...)

Great work by the way!

thanks!

[bhalonen]

Yeah, more or less that's what I concluded... this is just much much higher quality and I'm just a beginner in GNNs.

I'm working on integrating this to AlphaZero.jl, so I may have bunch of questions, hope you have patience.

[bhalonen]

Batched operations very important, obviously. Basically a requirement.

[CarloLucibello]

I'm working on integrating this to AlphaZero.jl, so I may have bunch of questions, hope you have patience.

Cool, I'm very happy to provide any kind of support