Skip to content

Dataset · biglam

Historical European newspapers

biglam/europeana_newspapers

European newspaper texts from the 18th to early 20th century for training and research.

What's inside

The dataset is compiled from Europeana historical newspapers and processed by BigLAM. It contains around 32B tokens in several European languages, including German, French, Greek, Estonian, Finnish, and Croatian. Text is extracted from ALTO XML; modalities include images, tables, and text.

What you can do with it

  • Train a language model on historical texts
  • Verify OCR and post-correction of newspaper scans
  • Search for materials in digital humanities
Tasks
Text generation
Languages
German, French, Greek, Estonian, Finnish, Croatian, Yiddish, Polish and 4 more
Size
10M–100M rows
Downloads this month
2.4K
Likes
69

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("biglam/europeana_newspapers")