-
Notifications
You must be signed in to change notification settings - Fork 3.5k
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鈥檒l occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove size parameter from forward methods #3713
Comments
Yes, this is a great point. In general, we only need support for However, the commits indeed suggest that this is not consistent across all PyG operators, so I'm happy to refactor this :) |
Cool! Reading from the documentation, currently it seems that the following convolutional layers still use the
As I only recently took an interest towards PyG's technical internals, it would be better to have a list of layers that are to be refactored, leaving the rest as they are. |
Thank you! I just checked and it looks like only |
Of course, that's why I opened the issue :). |
馃洜 Proposed Refactor
Modify all
forward
methods not to referencesize
in their signature and arguments, and modify the tests to pass accordingly.Motivation
Taking hint from e42ac59 and 6c79652 it seems that the
size
parameter of theforward
function is being deprecated, as calling without passing any value still results in the correct propagation of tensors.Pitch
To the best of my knowledge, there is no need to let the user explicitly set the
size
, as already many networks can safely infer it based on input values.Additional context
I'm unsure whether this refactoring can apply to all layers. Please let me know if there are specific use cases where it is best to leave
size
in the signature.The text was updated successfully, but these errors were encountered: