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

Trigger RuntimeError for when normalization is requested but not utilized #76

Draft
wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Trigger RuntimeError for when normalization is requested but not utilized #76

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
55dc569
refactor: making base task _compute_losses recognize norm
laserkelvin Dec 8, 2023
6e89424
feat: include additional number of keys check
laserkelvin Dec 8, 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
14 changes: 14 additions & 0 deletions matsciml/models/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -894,12 +894,26 @@ def _compute_losses(
targets = self._get_targets(batch)
predictions = self(batch)
losses = {}
used_norm = []
for key in self.task_keys:
target_val = targets[key]
if self.uses_normalizers:
target_val = self.normalizers[key].norm(target_val)
used_norm.append(key)
losses[key] = self.loss_func(predictions[key], target_val)
total_loss: torch.Tensor = sum(losses.values())
# trigger warning for when we infer normalization intent but
# not actually executed
if len(used_norm) == 0 and self.uses_normalizers:
raise RuntimeError(
"Target normalization was intended but not used."
f"Please check your config - expected: {self.task_keys}"
)
if len(used_norm) != len(self.normalizers):
raise RuntimeError(
"Normalization was performed, but number of keys do not match."
f"Expected {len(self.normalizers)} keys, but only used {len(used_norm)}."
)
return {"loss": total_loss, "log": losses}

def configure_optimizers(self) -> torch.optim.AdamW:
Expand Down