Removing episodes from LeRobotDataset

[andlyu]

Hi, thanks for building this. It's great.

Is there a way to easily remove episodes from a dataset. I had a decent amount of diversity in my episodes, and wanted to reduce it, so I had to remove ~1/2 of the episodes. Rather than rerecording them, I wanted to remove specified episodes (lets say all even episodes). Is there an easy way to do this? I'de tried just removing them from the episodes.jsonl file, but it seemed to load all of the episodes, and also deleting unwated episode videos/data and renaming the files through some issues when loading the datasets. Is there a better way to do this?

[The-Michael-Chen]

I am also wondering if there is a quick way to remove specific episodes from a dataset?

[andlyu]

I'de seen it's possible to do somethign like this:

        dataset = LeRobotDataset(
            cfg.repo_id, # username/dataset_name
            root=cfg.root, # Can be None
            episodes=range(0,9)
        )

[tlpss]

@andlyu how did you end up doing this?

[tlpss]

FYI: implemented in #831

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity (6 months). It will be closed if no further activity occurs. Thank you for your contributions.

[jadechoghari]

This was fixed with lerobot dataset edit tools: https://huggingface.co/docs/lerobot/en/using_dataset_tools#delete-episodes