groupfairnessllm/race_bias_with_human_label

这是一个用于测量大型语言模型长篇回答中种族相关偏见的人类标注数据集。每个示例包含一个参考回答和两个候选回答，这些回答都是对包含明确年龄提示的问题的回应。标注者根据提供的评估模板，判断哪个候选回答在语义上更接近参考回答。

This is a human-labeled dataset for measuring race-related bias in long-form responses generated by large language models. Each example includes a reference response and two candidate responses to a prompt with explicit age cues. Annotators determine which candidate response is semantically closer to the reference response using the provided evaluation template.