You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Allow loading tasks without downloading split files #1245
I analyzed the codebase and found that download_splits parameter already exists in method get_task(), and defaults to false in this PR #1345 in commit 2581671 by @PGijsbers
Should this issue be considered as completed, or is there additional work needed to provide some true lazy loading?
The only thing I noticed is that the method get_train_test_split_indices (line 154)
, downloads split files automatically; it doesn't respect the user's intent to avoid downloading splits and will fetch them even when download_splits=False.
Hi @mfeurer! 👋
As @OmarAI2003 has verified that download_splits parameter already exists in method get_task(), and defaults to false.
So should the issue be closed or is there any other task to be done under this issue.
Please calarify that so we can work on it.
Activity
OmarAI2003 commentedon Nov 15, 2025
I'm interested in this issue and plan to explore it in the coming days.
OmarAI2003 commentedon Nov 21, 2025
Hi @mfeurer! 👋
I analyzed the codebase and found that
download_splitsparameter already exists in methodget_task(), and defaults tofalsein this PR #1345 in commit 2581671 by @PGijsbersShould this issue be considered as completed, or is there additional work needed to provide some true lazy loading?
The only thing I noticed is that the method get_train_test_split_indices (line 154)
openml-python/openml/tasks/task.py
Line 154 in 4b1bdf4
AtharvaManale commentedon Oct 5, 2026
Hi @mfeurer! 👋
As @OmarAI2003 has verified that
download_splitsparameter already exists in methodget_task(),and defaults tofalse.So should the issue be closed or is there any other task to be done under this issue.
Please calarify that so we can work on it.
PGijsbers commentedon Oct 5, 2026
Yep, thanks for catching this!