STREAM

该数据集是一份由志愿者和多种流行的大型语言模型（LLM）注释的道德情境与道德判断数据的全面集合。它详细展示了在多种道德背景下，人类与人工智能的道德偏好和表现。该数据集包含了覆盖不同道德情境的多样化伦理场景，并由人类志愿者和LLM的混合注释构成，旨在助力AI模型与人类道德价值观的对齐。这是一个大规模且持续扩展的数据集，其任务是对道德判断进行评估以及生成伦理场景。

This dataset is a comprehensive collection of moral scenarios and moral judgment data annotated by volunteers and a variety of popular large language models (LLMs). It provides a detailed portrayal of the moral preferences and performance of humans and artificial intelligence across diverse moral contexts. The dataset includes diverse ethical scenarios spanning a wide range of moral situations, and is composed of mixed annotations from human volunteers and LLMs, aiming to facilitate the alignment between AI models and human moral values. This is a large-scale and continuously expanding dataset, with its core tasks being the evaluation of moral judgments and the generation of ethical scenarios.