@inproceedings{aynetdinov-etal-2026-repetition,
title = "Repetition over Diversity: High-Signal Data Filtering for Sample-Efficient {G}erman Language Modeling",
author = "Aynetdinov, Ansar and
Haller, Patrick and
Akbik, Alan",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1473/",
pages = "33690--33710",
ISBN = "979-8-89176-459-0",
abstract = "Recent research has shown that filtering massive English web corpora into high-quality subsets significantly improves training efficiency. However, for high-resource non-English languages like German, French, or Japanese, aggressive filtering creates a strategic dilemma: should practitioners prioritize *diversity* by training once on large amounts of lightly filtered web data, or prioritize *quality* by strictly filtering for a high-quality core and repeating it over multiple epochs?\par We investigate this trade-off for German by applying hierarchical quality filters to 500M web documents, comparing multi-epoch training on filtered subsets against single-pass training on a diverse corpus. Our experiments across multiple model scales and token budgets show that repeating high-quality data consistently outperforms single-pass training on larger, less filtered sets. Notably, the performance gap persists even after 7 epochs. Our findings suggest that for non-English LLMs, semantic concentration through quality filtering offers a more viable path to efficient language modeling than simply maximizing unique data volume.\par We release our German language models (called **Boldt**), as well as our cleaned evaluation benchmarks to the research community. Our experiments indicate that they achieve state-of-the-art results despite training on 10-360x fewer tokens than comparable models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="aynetdinov-etal-2026-repetition">
<titleInfo>
<title>Repetition over Diversity: High-Signal Data Filtering for Sample-Efficient German Language Modeling</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ansar</namePart>
<namePart type="family">Aynetdinov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Patrick</namePart>
<namePart type="family">Haller</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Akbik</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Recent research has shown that filtering massive English web corpora into high-quality subsets significantly improves training efficiency. However, for high-resource non-English languages like German, French, or Japanese, aggressive filtering creates a strategic dilemma: should practitioners prioritize *diversity* by training once on large amounts of lightly filtered web data, or prioritize *quality* by strictly filtering for a high-quality core and repeating it over multiple epochs?\par We investigate this trade-off for German by applying hierarchical quality filters to 500M web documents, comparing multi-epoch training on filtered subsets against single-pass training on a diverse corpus. Our experiments across multiple model scales and token budgets show that repeating high-quality data consistently outperforms single-pass training on larger, less filtered sets. Notably, the performance gap persists even after 7 epochs. Our findings suggest that for non-English LLMs, semantic concentration through quality filtering offers a more viable path to efficient language modeling than simply maximizing unique data volume.\par We release our German language models (called **Boldt**), as well as our cleaned evaluation benchmarks to the research community. Our experiments indicate that they achieve state-of-the-art results despite training on 10-360x fewer tokens than comparable models.</abstract>
<identifier type="citekey">aynetdinov-etal-2026-repetition</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1473/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>33690</start>
<end>33710</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Repetition over Diversity: High-Signal Data Filtering for Sample-Efficient German Language Modeling
%A Aynetdinov, Ansar
%A Haller, Patrick
%A Akbik, Alan
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F aynetdinov-etal-2026-repetition
%X Recent research has shown that filtering massive English web corpora into high-quality subsets significantly improves training efficiency. However, for high-resource non-English languages like German, French, or Japanese, aggressive filtering creates a strategic dilemma: should practitioners prioritize *diversity* by training once on large amounts of lightly filtered web data, or prioritize *quality* by strictly filtering for a high-quality core and repeating it over multiple epochs?\par We investigate this trade-off for German by applying hierarchical quality filters to 500M web documents, comparing multi-epoch training on filtered subsets against single-pass training on a diverse corpus. Our experiments across multiple model scales and token budgets show that repeating high-quality data consistently outperforms single-pass training on larger, less filtered sets. Notably, the performance gap persists even after 7 epochs. Our findings suggest that for non-English LLMs, semantic concentration through quality filtering offers a more viable path to efficient language modeling than simply maximizing unique data volume.\par We release our German language models (called **Boldt**), as well as our cleaned evaluation benchmarks to the research community. Our experiments indicate that they achieve state-of-the-art results despite training on 10-360x fewer tokens than comparable models.
%U https://aclanthology.org/2026.emnlp-main.1473/
%P 33690-33710
Markdown (Informal)
[Repetition over Diversity: High-Signal Data Filtering for Sample-Efficient German Language Modeling](https://aclanthology.org/2026.emnlp-main.1473/) (Aynetdinov et al., EMNLP 2026)
ACL