li-lab/HealthBenchX
收藏Hugging Face2025-09-29 更新2025-10-18 收录
下载链接：
https://hf-mirror.com/datasets/li-lab/HealthBenchX
下载链接
链接失效反馈官方服务：
资源简介：
HealthBenchXP是基于GPT-5机器翻译的HealthBench对话的版本，包含5000个现实中的健康对话，这些对话由跨越60个国家的262名医生合作构建，并配备了定制的医生创建的评价量表。该数据集的目的是为了在日语环境中评估医疗语言模型的能力。
HealthBenchXP is a machine-translated version of HealthBench conversations based on GPT-5, containing 5,000 realistic health conversations constructed by 262 physicians across 60 countries, each paired with a custom physician-created rubric for evaluation. The dataset aims to assess the capabilities of medical language models in Japanese contexts.
提供机构：
li-lab
