Model does not converge when running model_tutorial.ipynb (loss increases after training)

[zz310]

Hi, thanks for releasing this great project!

I was trying to reproduce the results by running model_tutorial.ipynb following the default settings.
However, I noticed that after 20 epochs of training, the model does not converge — in fact, the training loss slightly increases.

• Initial training loss: ~0.47
• Final training loss (after 20 epochs): ~0.51

Also, When I load the Norman dataset, I see multiple warnings like:
These perturbation are not in the GO graph and their perturbation can thus not be predicted.

Can I know if these warning are expected?
Thanks for your help!