---
license: cc-by-4.0
tags:
- curriculum
- huggingface
- lenient
- light
- minimal
- music
- pointcloud-text
- random-90-10
- weak-supervision
---

# prepare.py

## Dataset Summary

A **music** dataset with **pointcloud text** modality, stored in **huggingface** format.

## Preprocessing & Augmentation

- **Preprocessing**: minimal
- **Augmentation**: light

## Splits & Sampling

- **Split strategy**: random 90 10
- **Sampling**: curriculum

## Quality & Labeling

- **Quality filtering**: lenient
- **Labeling**: weak supervision

## Files

- `prepare.py` — main artifact of this repository

## License

See the license field above.