Dataset for Aligning Reasons (DFAR)

该数据集名为DFAR，精心挑选了5000个数据点，并按照9:1的比例划分为训练集和测试集。它旨在帮助语言模型生成类似人类的理由，以适应伦理决策。数据集中包含了伦理与不伦理的标签及其对应的原因，这被用来评估模型在伦理推理方面的表现。规模上，该数据集共有5000个数据点，其中4500个用于训练，500个用于测试。所涉及的任务是对伦理与非伦理进行分类并生成相应理由。

This dataset, named DFAR, consists of 5000 carefully curated data points, which are split into training and test sets at a 9:1 ratio. It is designed to help language models generate human-like rationales for ethical decision-making. The dataset contains both ethical and unethical labels along with their corresponding justifications, which are used to evaluate models' performance in ethical reasoning. A total of 4500 data points are allocated for training, while the remaining 500 are reserved for testing. The core task involved is to classify content as either ethical or unethical and generate corresponding justifications.