CliMedBench
收藏资源简介:
CliMedBench是一个大规模的中文医疗大语言模型评估基准,由华东师范大学等机构创建。该数据集包含33,735个问题,涵盖14个核心临床场景,主要来源于顶级三级医院的真实电子健康记录和考试练习。数据集的创建过程包括专家指导的数据选择和多轮质量控制,确保数据的真实性和可靠性。CliMedBench旨在评估和提升医疗大语言模型在临床决策支持、诊断和治疗建议等方面的能力,解决医疗领域中模型性能评估的不足问题。
CliMedBench is a large-scale Chinese medical large language model evaluation benchmark developed by institutions including East China Normal University. This dataset comprises 33,735 questions spanning 14 core clinical scenarios, primarily sourced from real electronic health records and practice examinations from top tertiary hospitals. The dataset construction process incorporates expert-guided data selection and multi-round quality control to guarantee the authenticity and reliability of the collected data. CliMedBench aims to evaluate and enhance the capabilities of medical large language models in clinical decision support, diagnosis and treatment recommendations, addressing the shortcomings of model performance assessment in the medical field.
CliMedBench




