You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for sharing the equiformer repo! I was wondering if there were any efforts towards creating an updated environment yaml file with newer versions of packages like PyTorch? The interface for some of the layers in PyTorch have changed between version 1.10.2 and the most recent version 2.4.0, such as the LayerNorm layer now accepting a keyword bias argument.
Thanks!
The text was updated successfully, but these errors were encountered:
Currently, I am not working on updating the environment.
Is there something specific that you want me to look at so that the codebase can be compatible with newer versions?
Regarding updating the environment, I think that it would just be good to upgrade everything so it works with the newest versions of PyTorch and PyTorch Lightning. We are currently trying to use the Equiformer as part of a larger framework which uses the newer versions of these packages and there have been some incompatibilities, most notably in the interface of layers like LayerNorm.
If possible, can you share the incompatibilities you have?
I think I will create a new branch and update the PyTorch version there. The main branch would still be the same since we want to make sure there is one copy of code that was used to reproduce the numbers claimed in the paper.
Hello,
Thanks for sharing the equiformer repo! I was wondering if there were any efforts towards creating an updated environment yaml file with newer versions of packages like PyTorch? The interface for some of the layers in PyTorch have changed between version 1.10.2 and the most recent version 2.4.0, such as the LayerNorm layer now accepting a keyword bias argument.
Thanks!
The text was updated successfully, but these errors were encountered: