Repetition over Diversity: High-Signal Data Filtering for Sample-Efficient German Language Modeling
For practitioners of non-English LLMs, this work demonstrates that quality filtering with repetition is more sample-efficient than maximizing data diversity, challenging the prevailing approach for high-resource languages.
The paper investigates the trade-off between data diversity and quality for German language modeling, finding that multi-epoch training on strictly filtered high-quality data outperforms single-pass training on larger, less filtered sets, achieving state-of-the-art results with 10-360x fewer tokens.
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? We investigate this trade-off for German by constructing hierarchical quality filters applied to 500M web documents, comparing multi-epoch training on the 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. 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.