Hungarian Document Datasets
Collection
Hungarian PDF pages from Common Crawl. Annotated with synthetic QAs by Llama 3.3 70B. • 3 items • Updated
Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
HuCCPDF
Approximately 113k pages of Hungarian PDFs from the Common Crawl, as featured in our paper "Synthetic Document Question Answering in Hungarian". The text field is extracted using PyMuPDF, and the ocr field is extracted using pytesseract.
See other datasets from the paper:
- HuDocVQA (https://huggingface.co/datasets/jlli/HuDocVQA)
- HuDocVQA-manual (https://huggingface.co/datasets/jlli/HuDocVQA-manual)
- Downloads last month
- 381