ChiKhaPo: A Large-Scale Multilingual Benchmark for Evaluating Lexical Comprehension and Generation in Large Language Models
This addresses the problem of evaluating basic linguistic skills in underrepresented languages for AI researchers, though it is incremental as it builds on existing lexicons and data.
The authors tackled the lack of lexical competence evaluation in low-resource languages for large language models by introducing ChiKhaPo, a benchmark covering 2700+ languages, and found that 6 state-of-the-art models performed poorly on it.
Existing benchmarks for large language models (LLMs) are largely restricted to high- or mid-resource languages, and often evaluate performance on higher-order tasks in reasoning and generation. However, plenty of evidence points to the fact that LLMs lack basic linguistic competence in the vast majority of the world's 3800+ written languages. We introduce ChiKhaPo, consisting of 8 subtasks of varying difficulty designed to evaluate the lexical comprehension and generation abilities of generative models. ChiKhaPo draws on existing lexicons, monolingual data, and bitext, and provides coverage for 2700+ languages for 2 subtasks, surpassing any existing benchmark in terms of language coverage. We further show that 6 SOTA models struggle on our benchmark, and discuss the factors contributing to performance scores, including language family, language resourcedness, task, and comprehension versus generation directions. With ChiKhaPo, we hope to enable and encourage the massively multilingual benchmarking of LLMs.