GNN nodes not updating

Hi all,
I’ve been running the code posted on the tutorial here: Training a GNN for Graph Classification — DGL 0.8.2post1 documentation but I don’t believe the node features are updating. I added print statements before and after the back propagation during the training loop occurred and the batched_graph.ndata[‘h’] values are the same. Is there something wrong with the code or my understanding?

Thanks!

hi @grapher, graph classification is a kind of inductive learning, which means it updates model parameters but not embed-dings, you can check the model params or valid metrics and will see the change.

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