Estimating Uncertainty from Reasoning: A Large-Scale Study of Multi- and Crosslingual MCQA Performance in LLMs

A large-scale study on the performance of multi- and cross-lingual multiple-choice questions (MCQA) for large language models (LLMs) aims to estimate the uncertainty of the models during reasoning processes. Through extensive datasets and experiments, this study analyzes the differences in model performance and uncertainty characteristics across different linguistic contexts, providing empirical evidence for improving the reliability and interpretability of large models.