---
license: apache-2.0
tags:
- adaptive
- audio-text
- lenient
- mixup-cutmix
- npy-sharded
- pseudo-label
- random
- random-90-10
- sports
---

# loader.py

## Dataset Summary

A **sports** dataset with **audio text** modality, stored in **npy sharded** format.

## Preprocessing & Augmentation

- **Preprocessing**: adaptive
- **Augmentation**: mixup cutmix

## Splits & Sampling

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

## Quality & Labeling

- **Quality filtering**: lenient
- **Labeling**: pseudo label

## Files

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

## License

See the license field above.