Klaudia Thellmann

Mehdi Ali, Michael Fromm, Klaudia Thellmann et al.

The recent success of Large Language Models (LLMs) has been predominantly driven by curating the training dataset composition, scaling of model architectures and dataset sizes and advancements in pretraining objectives, leaving tokenizer influence as a blind spot. Shedding light on this underexplored area, we conduct a comprehensive study on the influence of tokenizer choice on LLM downstream performance by training 24 mono- and multilingual LLMs at a 2.6B parameter scale, ablating different tokenizer algorithms and parameterizations. Our studies highlight that the tokenizer choice can significantly impact the model's downstream performance and training costs. In particular, we find that the common tokenizer evaluation metrics fertility and parity are not always predictive of model downstream performance, rendering these metrics a questionable proxy for the model's downstream performance. Furthermore, we show that multilingual tokenizers trained on the five most frequent European languages require vocabulary size increases of factor three in comparison to English. While English-centric tokenizers have been applied to the training of multi-lingual LLMs in the past, we find that this approach results in a severe downstream performance degradation and additional training costs of up to 68%, due to an inefficient tokenization vocabulary.

Mehdi Ali, Michael Fromm, Klaudia Thellmann et al.

We present two multilingual LLMs, Teuken 7B-base and Teuken 7B-instruct, designed to embrace Europe's linguistic diversity by supporting all 24 official languages of the European Union. Trained on a dataset comprising around 60% non-English data and utilizing a custom multilingual tokenizer, our models address the limitations of existing LLMs that predominantly focus on English or a few high-resource languages. We detail the models' development principles, i.e., data composition, tokenizer optimization, and training methodologies. The models demonstrate strong performance across multilingual benchmarks, as evidenced by their performance on European versions of ARC, HellaSwag, and TruthfulQA.

Klaudia Thellmann, Bernhard Stadler, Michael Fromm et al.

The rise of Large Language Models (LLMs) has revolutionized natural language processing across numerous languages and tasks. However, evaluating LLM performance in a consistent and meaningful way across multiple European languages remains challenging, especially due to the scarcity of language-parallel multilingual benchmarks. We introduce a multilingual evaluation approach tailored for European languages. We employ translated versions of five widely-used benchmarks to assess the capabilities of 40 LLMs across 21 European languages. Our contributions include examining the effectiveness of translated benchmarks, assessing the impact of different translation services, and offering a multilingual evaluation framework for LLMs that includes newly created datasets: EU20-MMLU, EU20-HellaSwag, EU20-ARC, EU20-TruthfulQA, and EU20-GSM8K. The benchmarks and results are made publicly available to encourage further research in multilingual LLM evaluation.

Alexander Arno Weber, Klaudia Thellmann, Jan Ebert et al.

The adaption of multilingual pre-trained LLMs into eloquent and helpful assistants is essential to facilitate their use across different language regions. In that spirit, we are the first to conduct an extensive study of the performance of multilingual models instruction-tuned on different language compositions on parallel instruction-tuning benchmarks across a selection of the most spoken Indo-European languages. We systematically examine the effects of language and instruction dataset size on a mid-sized and a large, multilingual LLMs by instruction-tuning them on parallel instruction-tuning datasets. Our results demonstrate that instruction-tuning on parallel instead of monolingual corpora benefits cross-lingual instruction following capabilities by up to 9.9%. Furthermore, we show that the Superficial Alignment Hypothesis does not hold in general, as the investigated multilingual 7B parameter model presents a counter-example requiring large-scale instruction-tuning datasets. Finally, we conduct a human annotation study to understand the alignment between human-based and GPT-4-based evaluation within multilingual chat scenarios.

Klaudia Thellmann, Bernhard Stadler, Michael Färber

Machine-translated benchmark datasets reduce costs and offer scale, but noise, loss of structure, and uneven quality weaken confidence. What matters is not merely whether we can translate, but also whether we can measure and verify translation reliability at scale. We study translation quality in the EU20 benchmark suite, which comprises five established benchmarks translated into 20 languages, via a three-step automated quality assurance approach: (i) a structural corpus audit with targeted fixes; (ii) quality profiling using a neural metric (COMET, reference-free and reference-based) with translation service comparisons (DeepL / ChatGPT / Google); and (iii) an LLM-based span-level translation error landscape. Trends are consistent: datasets with lower COMET scores exhibit a higher share of accuracy/mistranslation errors at span level (notably HellaSwag; ARC is comparatively clean). Reference-based COMET on MMLU against human-edited samples points in the same direction. We release cleaned/corrected versions of the EU20 datasets, and code for reproducibility. In sum, automated quality assurance offers practical, scalable indicators that help prioritize review -- complementing, not replacing, human gold standards.