-
Notifications
You must be signed in to change notification settings - Fork 88
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ideas on Message Passing #2
Comments
Hi there, thanks for your question. This question is mainly about how haiku works. The answer may be different depending on if you switch to another NN framework like flax, but the good news is that you can carry on using Jraph 👍 Haiku nets must be wrapped in another function. For example, from their docs:
So in your case you would need to write your
It's convenient to write the whole graph net inside a function, that way you only need to apply haikus
You can then use Not: if you are using a configured graph net, by default you will also receive global features, and features from the edges for which you node is a sender. So just for completeness, you will have to handle those.
Hope that's helpful, if you have any more questions let me know! |
I) I'm looking onto implementing the convolution operation specified in Graph Isomorphism Network.
Which transforms the Node Features by a set of Dense Layers. Can an haiku.module object be called on the node transformation function? If not how should that be done?
The text was updated successfully, but these errors were encountered: