Historisches Grundbuch der Stadt Basel Language Model

These are two (forward & backward) character-level language models, based on the Historical Land Records of the City of Basel.

Model Details

• Model Type: Contextual Character-level Language Model (Forward/Backward)
• Framework: Flair
• Language: German, Latin
• Era: 19th/20th century regests, based on original documents from 13.-18. c.

Training Data

This model was finetuned on the full HGB corpus. The training data comprises approximately 120,000 regests dating up to the year 1800. Please note the regests were written later, during the 19th and 20th century. The similarity to the original documents that they are based on varies strongly. As base model, we used the modern German Flair models, de-forward and de-backward.

Training Procedure

The calculations and model training were performed on UBELIX, the High-Performance Computing (HPC) cluster at the University of Bern.

Citation

This model was trained by Ismail Prada Ziegler as part of the Economies of Space project.

If you use this model in your research, please cite:

Prada Ziegler, Ismail. (2026). HGB Historical Language Model. Hugging Face. https://huggingface.co/dh-unibe/hgb-lm-v1