SOCIAL CHEMISTRY 101
收藏资源简介:
SOCIAL CHEMISTRY 101数据集由四川大学开发,包含292,000个社会规范的详细描述,每个规范包括一个行为和多个属性,如法律或非法、好或坏等人类判断。该数据集用于评估大型语言模型在处理社会规范时的毒性、偏见和价值对齐情况。通过使用此数据集,研究者能够更全面地理解语言模型在伦理方面的表现,并推动开发更加符合人类价值观和社会责任感的语言模型。
The SOCIAL CHEMISTRY 101 dataset was developed by Sichuan University. It contains 292,000 detailed descriptions of social norms, where each norm includes a behavior and multiple attributes reflecting human judgments such as whether a behavior is legal or illegal, good or bad. This dataset is used to evaluate the toxicity, bias, and value alignment of large language models (LLMs) when processing social norms. By leveraging this dataset, researchers can gain a more comprehensive understanding of the ethical performance of language models, and promote the development of language models that better align with human values and social responsibility.




