About GlobalAttentionPooling

I’ve noticed that parameter “gate_nn” is a neural network that computes attention scores for each feature. However , the return values do not include attention scores.
I’m wondering if there is any explanation?

I think it’s a great idea to allow optionally returning the attention values as in GATConv. Could you open a PR and add this functionality?

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.