IRLBench

该数据集名为IRLBench，是一个由2024年爱尔兰中学毕业考试中的12个代表性科目发展而来的基准测试集。它旨在评估模型在英语和爱尔兰语中的长篇生成能力。该基准测试支持对正确性和语言保真度的全面评估，尤其是在资源较少的语言如爱尔兰语中。该数据集涵盖了12个科目，任务是对开放性推理能力的评估。

This dataset, named IRLBench, is a benchmark collection developed from 12 representative subjects of the 2024 Irish Leaving Certificate Examination. It aims to evaluate the long-form generation capabilities of models in both English and Irish. This benchmark supports comprehensive evaluations of correctness and linguistic fidelity, especially in low-resource languages such as Irish. The dataset covers 12 subjects, with tasks targeting the assessment of open-ended reasoning abilities.