Skip to content
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

Frame Averaging Energy Reduction Bugfix #205

Merged

Conversation

melo-gonzo
Copy link
Collaborator

Currently, when frame averaging is uses, the predicted energy values are stored in a list because each frame will have an associated prediction. When reduced, this would collapse all of the predictions into a single scalar value, instead of a scalar value per sample in a batch. This PR addresses this issue by concatenating all predicted energy outputs prior to the reduction.

This PR also adds tests for FAENet for this specific bug as well as general forward pass checks for the applicable datasets.

This PR adds a check to keys in a batch, and includes natoms if not present, inside of the read_batch of FAENet itself.

@melo-gonzo melo-gonzo requested a review from laserkelvin May 3, 2024 22:02
@laserkelvin
Copy link
Collaborator

This doesn't seem to have the stuff from #204 - can you rebase?

matsciml/models/pyg/faenet/faenet.py Show resolved Hide resolved
matsciml/models/base.py Show resolved Hide resolved
@laserkelvin laserkelvin merged commit 4f6bdcb into IntelLabs:main May 6, 2024
2 of 3 checks passed
@melo-gonzo melo-gonzo deleted the faenet-bugfix-energy-reduction branch June 11, 2024 16:57
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants