Create a specific mask (no train nor test)

[AlejandroTL]

Hello!

First of all: this package is simply amazing, thanks.

My doubt is how can I create a specific mask for the huge graph that I have. In the pipeline I'm currently using I have a huge graph but I want to input, at different parts of the model, two subgraphs of the graph.
The first imputation is the huge graph with the exception of a few nodes, the second part is a much smaller subgraph that involves the previously ignored nodes. I assume that the best way of doing it is using a mask (or two masks, actually) over the huge graph that I have already build. I don't want these masks to be train masks or whatever, is just the way in which I think its easier to generate the subgraphs that I want, instead of generating them from scratch given that they share same topology and features.

Thanks!

[wsad1]

subgraph can you call this twice, with different masks. It creates a new edge_index so you don't have to create features from scratch.