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

Fix: 'IterableDatasetWrapper' has no len() when using Lhotse datasets #12190

Draft
wants to merge 3 commits into
base: main
Choose a base branch
from
Draft

Fix: 'IterableDatasetWrapper' has no len() when using Lhotse datasets #12190

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
296c51f
fix: lhotse has no len
AudranBert Feb 14, 2025
ac0a196
format file
AudranBert Feb 14, 2025
c78e0b9
Fix tarred datasets
AudranBert Feb 21, 2025
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
12 changes: 11 additions & 1 deletion nemo/collections/asr/models/ctc_models.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -413,11 +413,21 @@ def setup_training_data(self, train_data_config: Optional[Union[DictConfig, Dict
# We also need to check if limit_train_batches is already set.
# If it's an int, we assume that the user has set it to something sane, i.e. <= # training batches,
# and don't change it. Otherwise, adjust batches accordingly if it's a float (including 1.0).
if self._trainer is not None and isinstance(self._trainer.limit_train_batches, float):
if (
self._trainer is not None
and isinstance(self._trainer.limit_train_batches, float)
and (not hasattr(self._train_dl.dataset, 'dataset') or not isinstance(self._train_dl.dataset.dataset, LhotseSpeechToTextBpeDataset))
):
self._trainer.limit_train_batches = int(
self._trainer.limit_train_batches
* ceil((len(self._train_dl.dataset) / self.world_size) / train_data_config['batch_size'])
)
elif (
self._trainer is not None
and isinstance(self._trainer.limit_train_batches, float)
and (hasattr(self._train_dl.dataset, 'dataset') and isinstance(self._train_dl.dataset.dataset, LhotseSpeechToTextBpeDataset))
):
logging.warning(f"Lhotse dataset don't have length attribute, limit_train_batches is ignored.")
elif self._trainer is None:
logging.warning(
"Model Trainer was not set before constructing the dataset, incorrect number of "
Expand Down
Loading