Bilingual Evaluation of Language Models on General Knowledge in University Entrance Exams with Minimal Contamination

TL;DR

This paper introduces UNED-ACCESS 2024, a bilingual dataset of university entrance exam questions in Spanish and English, and evaluates various models' performance, revealing insights into language and size effects on reasoning tasks.

Contribution

It provides a novel, publicly unreleased bilingual dataset and benchmarks multiple models, highlighting differences in performance across languages and model sizes.

Findings

Reasoning questions are challenging for models.

Smaller models perform worse and degrade faster in Spanish.

Performance gap between languages is minimal for top models.

Abstract

In this article we present UNED-ACCESS 2024, a bilingual dataset that consists of 1003 multiple-choice questions of university entrance level exams in Spanish and English. Questions are originally formulated in Spanish and translated manually into English, and have not ever been publicly released. A selection of current open-source and proprietary models are evaluated in a uniform zero-shot experimental setting both on the UNED-ACCESS 2024 dataset and on an equivalent subset of MMLU questions. Results show that (i) reasoning questions are challenging for models, (ii) smaller models perform worse than larger models and degrade faster in Spanish than in English and (iii) the performance gap between languages is negligible for the best models and grows up to 37% for smaller models. Model ranking on UNED-ACCESS 2024 is almost identical in English and Spanish, and has also a high correlation (0.98 Pearson) with ranking on MMLU, suggesting that a small dataset is sufficiently diverse and representative to measure performance by discipline.