@inproceedings{burns-etal-2026-aleph,
title = "Aleph-Alpha-{G}erman{W}eb: Improving {G}erman-language {LLM} pre-training with model-based data curation and synthetic data generation",
author = {Burns, Thomas F and
Parcalabescu, Letitia and
Waeldchen, Stephan and
Barlow, Michael and
Ziegltrum, Gregor and
Stampa, Volker and
Harren, Bastian and
Deiseroth, Bj{\"o}rn},
editor = "Demberg, Vera and
Inui, Kentaro and
Marquez, Llu{\'i}s",
booktitle = "Proceedings of the 19th Conference of the {E}uropean Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = mar,
year = "2026",
address = "Rabat, Morocco",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.eacl-long.58/",
pages = "1267--1283",
ISBN = "979-8-89176-380-7",
abstract = "Scaling data quantity is essential for large language models (LLMs), yet recent findings show that data quality can significantly boost performance and training efficiency. We introduce a German-language dataset curation pipeline that combines heuristic and model-based filtering techniques with synthetic data generation. We use our pipeline to create Aleph-Alpha-GermanWeb, a large-scale German pre-training dataset which draws from: (1) Common Crawl web data, (2) FineWeb2, and (3) synthetically-generated data conditioned on actual, organic web data. We evaluate our dataset by pre-training both a 1B Llama-style model and an 8B tokeniser-free hierarchical autoregressive transformer (HAT). A comparison on German-language benchmarks, including MMMLU, shows significant performance gains of Aleph-Alpha-GermanWeb over FineWeb2 alone. This advantage holds at the 8B scale even when FineWeb2 is enriched by human-curated high-quality data sources such as Wikipedia. Our findings support the growing body of evidence that model-based data curation and synthetic data generation can significantly enhance LLM pre-training datasets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="burns-etal-2026-aleph">
<titleInfo>
<title>Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Thomas</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Burns</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Letitia</namePart>
<namePart type="family">Parcalabescu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Stephan</namePart>
<namePart type="family">Waeldchen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michael</namePart>
<namePart type="family">Barlow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gregor</namePart>
<namePart type="family">Ziegltrum</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Volker</namePart>
<namePart type="family">Stampa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bastian</namePart>
<namePart type="family">Harren</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Björn</namePart>
<namePart type="family">Deiseroth</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-03</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Vera</namePart>
<namePart type="family">Demberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lluís</namePart>
<namePart type="family">Marquez</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Rabat, Morocco</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-380-7</identifier>
</relatedItem>
<abstract>Scaling data quantity is essential for large language models (LLMs), yet recent findings show that data quality can significantly boost performance and training efficiency. We introduce a German-language dataset curation pipeline that combines heuristic and model-based filtering techniques with synthetic data generation. We use our pipeline to create Aleph-Alpha-GermanWeb, a large-scale German pre-training dataset which draws from: (1) Common Crawl web data, (2) FineWeb2, and (3) synthetically-generated data conditioned on actual, organic web data. We evaluate our dataset by pre-training both a 1B Llama-style model and an 8B tokeniser-free hierarchical autoregressive transformer (HAT). A comparison on German-language benchmarks, including MMMLU, shows significant performance gains of Aleph-Alpha-GermanWeb over FineWeb2 alone. This advantage holds at the 8B scale even when FineWeb2 is enriched by human-curated high-quality data sources such as Wikipedia. Our findings support the growing body of evidence that model-based data curation and synthetic data generation can significantly enhance LLM pre-training datasets.</abstract>
<identifier type="citekey">burns-etal-2026-aleph</identifier>
<location>
<url>https://aclanthology.org/2026.eacl-long.58/</url>
</location>
<part>
<date>2026-03</date>
<extent unit="page">
<start>1267</start>
<end>1283</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation
%A Burns, Thomas F.
%A Parcalabescu, Letitia
%A Waeldchen, Stephan
%A Barlow, Michael
%A Ziegltrum, Gregor
%A Stampa, Volker
%A Harren, Bastian
%A Deiseroth, Björn
%Y Demberg, Vera
%Y Inui, Kentaro
%Y Marquez, Lluís
%S Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 March
%I Association for Computational Linguistics
%C Rabat, Morocco
%@ 979-8-89176-380-7
%F burns-etal-2026-aleph
%X Scaling data quantity is essential for large language models (LLMs), yet recent findings show that data quality can significantly boost performance and training efficiency. We introduce a German-language dataset curation pipeline that combines heuristic and model-based filtering techniques with synthetic data generation. We use our pipeline to create Aleph-Alpha-GermanWeb, a large-scale German pre-training dataset which draws from: (1) Common Crawl web data, (2) FineWeb2, and (3) synthetically-generated data conditioned on actual, organic web data. We evaluate our dataset by pre-training both a 1B Llama-style model and an 8B tokeniser-free hierarchical autoregressive transformer (HAT). A comparison on German-language benchmarks, including MMMLU, shows significant performance gains of Aleph-Alpha-GermanWeb over FineWeb2 alone. This advantage holds at the 8B scale even when FineWeb2 is enriched by human-curated high-quality data sources such as Wikipedia. Our findings support the growing body of evidence that model-based data curation and synthetic data generation can significantly enhance LLM pre-training datasets.
%U https://aclanthology.org/2026.eacl-long.58/
%P 1267-1283
Markdown (Informal)
[Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation](https://aclanthology.org/2026.eacl-long.58/) (Burns et al., EACL 2026)
ACL
- Thomas F Burns, Letitia Parcalabescu, Stephan Waeldchen, Michael Barlow, Gregor Ziegltrum, Volker Stampa, Bastian Harren, and Björn Deiseroth. 2026. Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1267–1283, Rabat, Morocco. Association for Computational Linguistics.