TUMLU: A Unified and Native Language Understanding Benchmark for Turkic Languages

TL;DR

This paper introduces TUMLU, a native, comprehensive benchmark for evaluating Turkic languages' understanding in AI models, addressing the lack of high-quality, culturally nuanced evaluation datasets for these under-represented languages.

Contribution

The paper presents TUMLU and TUMLU-mini benchmarks, the first native Turkic language understanding datasets, and evaluates various large language models on these benchmarks.

Findings

Open and proprietary LLMs show varied performance across Turkic languages.

TUMLU benchmarks reveal strengths and weaknesses of models in different linguistic and cultural contexts.

Release of datasets and evaluation scripts to foster further research.

Abstract

Being able to thoroughly assess massive multi-task language understanding (MMLU) capabilities is essential for advancing the applicability of multilingual language models. However, preparing such benchmarks in high quality native language is often costly and therefore limits the representativeness of evaluation datasets. While recent efforts focused on building more inclusive MMLU benchmarks, these are conventionally built using machine translation from high-resource languages, which may introduce errors and fail to account for the linguistic and cultural intricacies of the target languages. In this paper, we address the lack of native language MMLU benchmark especially in the under-represented Turkic language family with distinct morphosyntactic and cultural characteristics. We propose two benchmarks for Turkic language MMLU: TUMLU is a comprehensive, multilingual, and natively developed language understanding benchmark specifically designed for Turkic languages. It consists of middle- and high-school level questions spanning 11 academic subjects in Azerbaijani, Crimean Tatar, Karakalpak, Kazakh, Tatar, Turkish, Uyghur, and Uzbek. We also present TUMLU-mini, a more concise, balanced, and manually verified subset of the dataset. Using this dataset, we systematically evaluate a diverse range of open and proprietary multilingual large language models (LLMs), including Claude, Gemini, GPT, and LLaMA, offering an in-depth analysis of their performance across different languages, subjects, and alphabets. To promote further research and development in multilingual language understanding, we release TUMLU-mini and all corresponding evaluation scripts.