KoMT-Bench
收藏arXiv2025-09-30 收录
下载链接：
https://huggingface.co/datasets/LGAI-EXAONE/KoMT-Bench
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为KoMT-Bench，是我们内部制作的一套数据集，它通过对原始的MT-Bench数据集进行韩文翻译，并对内容进行修改，以体现韩语的特性和文化细微差别。其问题和类别的数量与原始的MT-Bench数据集保持一致。该数据集的任务是评估韩语语言模型的性能。
This dataset, named KoMT-Bench, is an in-house developed benchmark. It is constructed by translating the original MT-Bench dataset into Korean and revising the content to reflect the linguistic characteristics and cultural nuances of the Korean language. The number of questions and categories in KoMT-Bench is consistent with that of the original MT-Bench dataset. The primary task of this dataset is to evaluate the performance of Korean language models.
提供机构：
LG AI Research
