The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
Historical Basque newspapers OCR
Manually corrected transcriptions of historical Basque-language newspaper pages (Escualdun Gazeta, Eskual-Herria, Euzkadi, Argia; 1885-1923).
full_page: one sample per page — the full page scan and its full transcription (sections separated by a blank line).crops: one sample per page region — the cropped image and the transcription of that region.
Line breaks follow the original print (hyphenated words are not joined).
from datasets import load_dataset
ds = load_dataset("<user>/<repo>", "crops")
- Downloads last month
- 42