Datasets:
The dataset viewer is not available for this split.
Error code: TooBigContentError
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.
German Commons - 134 Billion Tokens of Openly Licensed Text for German Language Models
A large collection of German-language text under open licenses for training German language models.
Version 1.1 (2026-11-13). This version corrects the license labels of nine sources, checks the license of every document of three further sources, and removes seven sources entirely as well as 3,824 documents without an open license. No text of any remaining document was changed. The paper describes version 1.0 (35.78 million documents, 154.56 billion tokens), so its numbers no longer match this release. See the changelog for every change. Version 1.0 remains available with
revision="v1.0".
- Datasheet: DATASHEET.md
- Changelog: CHANGELOG.md
- Paper (describes version 1.0): arxiv.org/abs/2510.13996
- Code: github.com/coral-nlp/llmdata
- Bloom Filter (DOLMA-compatible, built for version 1.0): bloom_filter.bin
Dataset Description
This dataset is aggregated from 34 sources and contains 134.44 billion tokens of German text in 20.11 million documents, spanning 7 thematic domains:
- 🌐 Web Commons: 2.95B tokens sourced from Wikipedia
- 💬 Political Commons: 3.57B tokens sourced from parliamentary documents, speeches, protocols, and EU documents
- ⚖️ Legal Commons: 2.99B tokens sourced from court decisions, federal law, legal databases, EU legal documents
- 📰 News Commons: 72.67B tokens sourced from historical and current newspaper archives
- 🏦 Economics Commons: 0.11B tokens sourced from EU public tenders
- 📚 Cultural Commons: 51.34B tokens sourced from cultural heritage collections, Wiki projects, and speech transcripts
- 🔬 Scientific Commons: 0.80B tokens sourced from scholarly papers, books, and technical journals
Dataset Features
Each record contains the following fields:
- id: Identifier taken from the source dataset (not guaranteed to be unique)
- source: Source dataset name
- subset: Thematic subset (Cultural, Legal, Political, Scientific, News, Web, Economic)
- text: Main text content; deduplicated, quality filtered, with consistent formatting and encoding. Can be split at newlines to obtain paragraph text.
- license: List of applicable licenses for the document, given as canonical SPDX license URLs (
https://spdx.org/licenses/<ID>). In version 1.1, every document carries exactly one.CC-PDM-1.0is the Creative Commons Public Domain Mark: it marks a work as free of known copyright and is not a license. - num_tokens: GPT-2 token count
- perplexity: Text perplexity measured with a KenLM model trained on German Wikipedia text
- ocr_score: OCR quality score measured using OCRoscope
Dataset Usage
Load the entire dataset
from datasets import load_dataset ds = load_dataset("coral-nlp/german-commons")Load a thematic subset
ds = load_dataset("coral-nlp/german-commons", "cultural")Load individual source datasets
wikipedia = load_dataset("coral-nlp/german-commons", "web", split="wikipedia")Load version 1.0, including the sources removed in version 1.1
ds = load_dataset("coral-nlp/german-commons", revision="v1.0")
Supported splits and constituent datasets are:
| Subset | Split Key | Dataset Name | Docs | Tokens | License | Text Type | Source |
|---|---|---|---|---|---|---|---|
web |
wikipedia |
Wikipedia | 2,930,224 | 2,948,751,608 | CC-BY-SA-4.0 | Various | 🔗 |
political |
reichtagsprotokolle |
Reichtagsprotokolle | 522 | 703,495,637 | CC-BY-SA-4.0 | Parliamentary Protocols | 🔗 |
political |
germanpoliticalspeeches |
German Political Speeches | 6,678 | 29,409,655 | CC-BY-SA-4.0 | Speech Transcripts | 🔗 |
political |
btdrucksachen |
Corpus der Drucksachen des Deutschen Bundestages | 3,017 | 528,769,669 | CC0-1.0 | Parliamentary Publications | 🔗 |
political |
btplenarprotokolle |
Corpus der Plenarprotokolle des Deutschen Bundestages | 1,833 | 316,034,708 | CC0-1.0 | Parliamentary Protocols | 🔗 |
political |
eurovoc |
EuroVoc | 245,838 | 1,988,111,462 | CC-BY-4.0 | Parliamentary Publications | 🔗 |
legal |
bundesrecht |
Corpus des Deutschen Bundesrechts | 3,217 | 1,004,294 | CC0-1.0 | German Federal Laws | 🔗 |
legal |
openlegaldata |
OpenLegalData | 249,909 | 1,915,956,613 | CC-PDM-1.0 | Court Decisions | 🔗 |
legal |
bfh |
Corpus der Entscheidungen des BFH | 10,885 | 67,791,931 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bgh20 |
Entscheidungen des BGH in Strafsachen des 20. Jhd. | 36,062 | 92,873,390 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bgh |
Corpus der Entscheidungen des BGH | 77,258 | 292,832,709 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bverfg |
Corpus der Entscheidungen des BVerfG | 8,028 | 39,503,223 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bpatg |
Corpus der Entscheidungen des BpatG | 30,705 | 185,099,188 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bverwg |
Corpus der Entscheidungen des BVerwG | 27,185 | 123,487,739 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bverfgaes |
Corpus der amtl. Entscheidungssammlung des BVerfG | 919 | 24,427,294 | CC0-1.0 | Court Decisions | 🔗 |
legal |
bag |
Corpus der Entscheidungen des BAG | 5,624 | 48,248,111 | CC0-1.0 | Court Decisions | 🔗 |
legal |
eurlex |
EurLEX | 64,934 | 201,263,562 | CC-BY-4.0 | European Union Laws | 🔗 |
news |
zeitungsportal |
Deutsches Zeitungsportal | 8,076,164 | 43,871,094,547 | Various¹ | News Articles | 🔗 |
news |
europeana |
Europeana Newspapers | 3,256,341 | 20,684,418,365 | CC-PDM-1.0 | News Articles | 🔗 |
news |
anno |
ANNO | 1,910,281 | 8,103,825,248 | CC-PDM-1.0 | News Articles | 🔗 |
news |
wikinews |
WikiNews | 23,266 | 14,222,520 | CC-BY-2.5 | News Articles | 🔗 |
economic |
tedeutenders |
TEDEUTenders | 57,214 | 110,611,112 | CC-BY-4.0 | Procurement Notices | 🔗 |
cultural |
dibilit |
DiBiLit-Korpus | 2,062 | 216,391,448 | CC-BY-SA-4.0 | Literature | 🔗 |
cultural |
dibiphil |
DiBiPhil-Korpus | 270 | 32,151,997 | CC-BY-SA-4.0 | Literature | 🔗 |
cultural |
wikisource |
Wikisource | 240,689 | 347,770,430 | CC-BY-SA-4.0 | Various | 🔗 |
cultural |
wikivoyage |
Wikivoyage | 20,370 | 42,025,478 | CC-BY-SA-4.0 | Travel | 🔗 |
cultural |
germanpd |
German-PD | 123,914 | 49,333,198,231 | CC-PDM-1.0 | Literature | 🔗 |
cultural |
blbooks |
BLBooks | 3,714 | 1,012,047,216 | CC-PDM-1.0 | Literature | 🔗 |
cultural |
mosel |
MOSEL | 2,605,569 | 358,514,283 | CC-BY-4.0 | Speech Transcripts | 🔗 |
scientific |
wikibooks |
Wikibooks | 27,292 | 50,434,996 | CC-BY-SA-4.0 | Educational Books | 🔗 |
scientific |
polyjournal |
Digitalisierung des Polytechnischen Journals | 346 | 180,257,799 | CC-BY-SA-4.0 | Scholarly Papers | 🔗 |
scientific |
doab |
Directory of Open Access Books | 1,890 | 163,432,647 | Various¹ | Scholarly Books | 🔗 |
scientific |
openalex |
OpenAlex | 43,958 | 380,433,196 | Various¹ | Scholarly Papers | 🔗 |
scientific |
wikiversity |
Wikiversity | 16,371 | 27,802,099 | CC-BY-SA-4.0 | Educational Content | 🔗 |
¹ The license was checked per document; see the changelog. Deutsches Zeitungsportal: CC-PDM-1.0 (8,050,317), CC0-1.0 (25,847). DOAB: CC-BY-4.0 (1,287), CC-BY-SA-4.0 (576), older CC BY and CC BY-SA versions (27). OpenAlex: CC-PDM-1.0 (39,456), CC-BY-4.0 (4,090), CC-BY-SA-4.0 (412).
Version 1.1 removed the splits cultural/sbbfulltexts, cultural/wikiquote, scientific/arxiv, web/onemillionposts, web/thestack, web/wikidiscussions, and web/youtubecommons. The web subset now contains only Wikipedia.
Projects Using German Commons
If you've built something with German Commons, feel free to open a PR to add it here!
| Project | Description |
|---|---|
| OpenEuroLLM Propella Annotations | Document-level annotations via propella-1-4b: content classification, quality, audience, safety, and geographic relevance (built on version 1.0) |
Citation
If you use this dataset, please cite the corresponding paper. The paper describes version 1.0.
@article{gienapp:2025d,
title = {{The German Commons -- 154 Billion Tokens of Openly Licensed Text for German Language Models}},
author = {Lukas Gienapp and
Christopher Schr\"oder and
Stefan Schweter and
Christopher Akiki and
Ferdinand Schlatt and
Arden Zimmermann and
Phillipe Gen\^et and
Martin Potthast},
year = 2025,
month = oct,
journal = {CoRR},
volume = {abs/2510.13996},
url = {https://arxiv.org/abs/2510.13996}
}
License
The compilation and its metadata are licensed under ODC-BY 1.0. Each document keeps its own license, given in its license field. Version 1.1 contains nine values:
| License | Documents | Conditions |
|---|---|---|
| CC-PDM-1.0 | 13,633,932 | None known; marks works as free of known copyright |
| CC-BY-SA-4.0 | 3,245,812 | Attribution; adaptations must be shared under the same license |
| CC-BY-4.0 | 2,978,932 | Attribution |
| CC0-1.0 | 230,580 | None |
| CC-BY-2.5 | 23,267 | Attribution |
| CC-BY-SA-3.0-DE | 14 | Attribution; adaptations must be shared under the same license |
| CC-BY-SA-3.0 | 6 | Attribution; adaptations must be shared under the same license |
| CC-BY-3.0 | 5 | Attribution |
| CC-BY-3.0-DE | 1 | Attribution |
All nine permit commercial use. Users are responsible for meeting the conditions of each document's license.
- Downloads last month
- 3,869