LinkNeighborLoader weighted sampling

[denadai2]

Dear community,

I'd like to implement LinkNeighborLoader with a weighted sampling, giving a probability to each edge or node. Is there a trick to do it with the current pyg/pyg-lib/torch-scatter? If no, could you help me implementing it? I am not sure what's the best way to start e.g in pyg-lib

thx

[rusty1s]

We are currently working on this with - should land pretty soon. Progress will be tracked in #7331

[denadai2]

That's a GREAT news! Eager to ear more about it!