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.

Sign up for GitHub

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

Jump to bottom

Support for using tag weights with L2MAE force loss #439

Merged
merged 2 commits into from
Jan 19, 2023
Merged

Support for using tag weights with L2MAE force loss #439

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f547417
Support for using tag weights with L2MAE force loss
abhshkdz Jan 11, 2023
de4af8e
Merge branch 'main' into l2mae-tagweights
abhshkdz Jan 11, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 20 additions & 13 deletions ocpmodels/trainers/forces_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -489,21 +489,28 @@ def _compute_loss(self, out, batch_list):
weight[batch_tags == 1] = tag_specific_weights[1]
weight[batch_tags == 2] = tag_specific_weights[2]

loss_force_list = torch.abs(out["forces"] - force_target)
train_loss_force_unnormalized = torch.sum(
loss_force_list * weight.view(-1, 1)
)
train_loss_force_normalizer = 3.0 * weight.sum()

# add up normalizer to obtain global normalizer
distutils.all_reduce(train_loss_force_normalizer)
if self.config["optim"].get("loss_force", "l2mae") == "l2mae":
dists = torch.norm(
out["forces"] - force_target, p=2, dim=-1
)
weighted_dists_sum = (dists * weight).sum()

# perform loss normalization before backprop
train_loss_force_normalized = train_loss_force_unnormalized * (
distutils.get_world_size() / train_loss_force_normalizer
)
loss.append(train_loss_force_normalized)
num_samples = out["forces"].shape[0]
num_samples = distutils.all_reduce(
num_samples, device=self.device
)
weighted_dists_sum = (
weighted_dists_sum
* distutils.get_world_size()
/ num_samples
)

force_mult = self.config["optim"].get(
"force_coefficient", 30
)
loss.append(force_mult * weighted_dists_sum)
else:
raise NotImplementedError
else:
# Force coefficient = 30 has been working well for us.
force_mult = self.config["optim"].get("force_coefficient", 30)
Expand Down