CEIA-RL/Safety-Preference-Energy

该数据集是一个用于自然语言处理任务的数据集，特别适用于对话系统或模型比较任务。数据集包含提示（prompt）、选择的回答（chosen）、拒绝的回答（rejected）等特征，每个特征都有详细的子字段描述。此外，数据集还包含元数据（metadata），提供了关于数据块的额外信息，如风险级别、技术背景、威胁类型等。数据集分为训练集，包含48,211个示例和约797MB的数据。

This dataset is designed for natural language processing (NLP) tasks, and is particularly suited for dialogue systems or model comparison tasks. It includes features such as prompt, chosen response, rejected response, among others, with detailed sub-field descriptions for each feature. Additionally, the dataset contains metadata that provides supplementary information about the data chunks, including risk level, technical background, threat type, and more. The dataset is split into a training set, which contains 48,211 instances and approximately 797 MB of data.