afrimedqa/afrimedqa_v2

AfriMed-QA v2数据集是一个包含15,000个泛非临床多样性的多项选择和开放式问题的开源数据集，旨在严格评估LLMs在非洲医疗保健领域的表现。数据集来源于16个国家超过600名临床和非临床贡献者，涵盖32个临床专业，其地理和临床多样性有助于对LLMs的准确性、事实性、幻觉、人口统计偏差、潜在危害、理解和回忆进行稳健的上下文评估，并提供了一个足够大的语料库，以便微调LLMs以减轻发现的偏差。

AfriMed-QA v2 creates a novel multispecialty open-source dataset of 15,000 pan-African clinically diverse multiple choice and open ended questions and answers to rigorously evaluate LLMs in African healthcare. Sourced from over 600 clinical and non-clinical contributors across 16 countries and covering 32 clinical specialties, the dataset’s geographical and clinical diversity facilitates robust contextual evaluation of LLMs for accuracy, factuality, hallucination, demographic bias, potential for harm, comprehension, and recall, and provides a sufficiently large corpus to fine-tune LLMs to mitigate biases discovered.