How to achieve training with limited data (Limited Scenes)?

How do you perform semantic segmentation on a limited point cloud? Do you only use a small amount of labeled data? How do you handle unlabeled data?

The issue names no file, test, or entry point. First inspect the repository to identify the documented training and data-handling entry points, then determine whether limited labeled point-cloud data and unlabeled data are supported. Done means providing clear, repository-specific guidance for both cases.

Written by the indexing model from the issue text.