using SVDKL with GNN #2281
Unanswered
alonsocampana
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi,
I'm trying to perform gaussian process regression using GNNs (implemented in torch geometric), and Gpytorch.
It's a single output regression problem. I've adapted the code in the tutorial, but the learning is extremely slow, and probably it's because my implementation has issues. I'm using the initial graph-features generated by the GNN as inducing points, and I suspect the issue could be related to this, but I don't know how to do it differently.
Any tip on how should I face this problem?
EDIT: My learning rate was very low, now it learns something in training, but in test all the predictions are almost the same.
Beta Was this translation helpful? Give feedback.
All reactions