Fix memory leak in LeRobotDataset._save_episode_table #1113
+8
−3
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix memory leak in LeRobotDataset
Description
This PR fixes a memory leak in the
LeRobotDataset
class that causes memory usage to grow continuously when processing multiple episodes sequentially.Issue
When converting data to LeRobot format and processing multiple episodes in sequence, memory usage increases continuously without being released. Through memory profiling, I identified the issue in the
_save_episode_table
method, specifically with how dataset objects are accumulated in memory.Solution
Modified the
_save_episode_table
method to:self.hf_dataset
to a minimal empty dataset after saving, instead of accumulating dataThis approach preserves all functionality while preventing memory accumulation.
Validation
Tested with multiple episodes conversion: