---
license: cc-by-4.0
tags:
- contrastive
- document-ocr
- huggingface
- lenient
- manual
- multimodal3
- progressive
- randaugment
- random-90-10
---

# dataset_152013975_document_ocr_multimodal3.py

## Dataset Summary

A **document ocr** dataset with **multimodal3** modality, stored in **huggingface** format.

## Preprocessing & Augmentation

- **Preprocessing**: progressive
- **Augmentation**: randaugment

## Splits & Sampling

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

## Quality & Labeling

- **Quality filtering**: lenient
- **Labeling**: manual

## Files

- `dataset_152013975_document_ocr_multimodal3.py` — main artifact of this repository

## License

See the license field above.