Skip to content

Fix state_dict loading for models trained with torch.compile #263

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

Open
inorganicwriter wants to merge 1 commit into
base: main
Choose a base branch
from
+9 −2
Open

Fix state_dict loading for models trained with torch.compile #263

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
11 changes: 9 additions & 2 deletions googlehydrology/evaluation/tester.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -163,9 +163,16 @@ def _load_weights(self, epoch: int = None):
weight_file = self._get_weight_file(epoch)

LOGGER.info(f'Using the model weights from {weight_file}')
self.model.load_state_dict(
torch.load(weight_file, map_location=self.device, weights_only=True)
state_dict = torch.load(
weight_file, map_location=self.device, weights_only=True
)
# Drop `_orig_mod.` prefix introduced by torch.compile when the model
# was trained with compile=True but is now loaded without compilation.
if any(k.startswith('_orig_mod.') for k in state_dict):
state_dict = {
k[len('_orig_mod.'):]: v for k, v in state_dict.items()
}
self.model.load_state_dict(state_dict)

def _get_dataset_all(self) -> Dataset:
"""Get dataset for all basins."""
Expand Down