david9dragon9/shp_translations

该数据集包含Stanford Human Preference (SHP)数据集中三个部分（askscience, explainlikeimfive, legaladvice）的翻译，支持英语、韩语、中文和泰语，用于训练领域不变性的奖励模型。数据集的翻译是通过No Language Left Behind (NLLB) 3.3 B 200模型完成的，主要应用于问题回答任务，并且与法律相关。This dataset includes translations of three splits (askscience, explainlikeimfive, legaladvice) from the Stanford Human Preference (SHP) dataset, supporting English, Korean, Chinese, and Thai languages, used for training domain-invariant reward models. The translations were performed using the No Language Left Behind (NLLB) 3.3 B 200 model, primarily focused on the question-answering task with a legal relevance tag.