aieng-lab/gentypes
收藏Hugging Face2025-02-04 更新2025-04-26 收录
下载链接：
https://hf-mirror.com/datasets/aieng-lab/gentypes
下载链接
链接失效反馈官方服务：
资源简介：
GENTYPES数据集包含500个短句，这些短句将一个由模板掩码[NAME]表示的名字与刻板印象相关联。每个条目都包含文本、性别和原因三个字段，性别字段表示与文本中的刻板印象更强相关的性别（男或女），原因字段则指定了性别偏见的十个原因分类之一。该数据集可用于评估语言模型的性别偏见。
The GENTYPES dataset consists of 500 short sentences that link a name represented by the template mask [NAME] to stereotypical associations. Each entry includes the fields text, gender, and reason, with the gender field indicating the gender more strongly associated with the stereotype in the text, and the reason field specifying one of ten categories for gender bias. The dataset can be used to assess the gender bias of language models.
提供机构：
aieng-lab
