The German Commons - 154 Billion Tokens of Openly Licensed Text for German Language Models
Lukas Gienapp, Christopher Schröder, Stefan Schweter, Christopher Akiki, Ferdinand Schlatt, Arden Zimmermann, Phillipe Genêt, Martin Potthast
TL;DR
The paper tackles the scarcity of openly licensed German training data for large language models by introducing the German Commons, the largest openly licensed German text collection to date (154.56B tokens across seven thematic domains). It details a rigorous sourcing strategy from 41 sources with explicit licenses, a multi-stage processing pipeline (text extraction, formatting, language filtering, quality checks, deduplication, PII removal, and license mapping) and domain-specific provenance, culminating in a Parquet dataset with per-document SPDX licenses. The work includes extensive corpus statistics, discusses limitations (notably historical bias and OCR artifacts), and provides open-source tooling (llmdata) to ensure reproducibility and extensibility, thereby enabling truly open German language model development. The dataset is funded by German and European research programs and is intended to support cross-domain German NLP research and open-model training, reducing licensing and governance barriers in non-English language AI.
Abstract
Large language model development relies on large-scale training corpora, yet most contain data of unclear licensing status, limiting the development of truly open models. This problem is exacerbated for non-English languages, where openly licensed text remains critically scarce. We introduce the German Commons, the largest collection of openly licensed German text to date. It compiles data from 41 sources across seven domains, encompassing legal, scientific, cultural, political, news, economic, and web text. Through systematic sourcing from established data providers with verifiable licensing, it yields 154.56 billion tokens of high-quality text for language model training. Our processing pipeline implements comprehensive quality filtering, deduplication, and text formatting fixes, ensuring consistent quality across heterogeneous text sources. All domain subsets feature licenses of at least CC-BY-SA 4.0 or equivalent, ensuring legal compliance for model training and redistribution. The German Commons therefore addresses the critical gap in openly licensed German pretraining data, and enables the development of truly open German language models. We also release code for corpus construction and data filtering tailored to German language text, rendering the German Commons fully reproducible and extensible.