RuMedBench

RuMedBench是由Sber人工智能实验室创建的俄罗斯医疗语言理解基准数据集，包含五个任务，覆盖分类、问答、自然语言推理和命名实体识别等。数据集基于四种医疗文本类型，包括门诊记录、临床笔记等，共计处理了数十万条文本。创建过程中，团队对数据进行了严格的隐私保护和伦理审查。该数据集主要用于评估和提高模型在医疗领域的语言处理能力，特别是在处理敏感医疗信息时的表现。

RuMedBench is a Russian medical language understanding benchmark dataset developed by Sber Artificial Intelligence Laboratory. It encompasses five tasks covering classification, question answering (QA), natural language inference (NLI), named entity recognition (NER) and other related tasks. The dataset is built on four categories of medical texts, including outpatient records, clinical notes and others, with hundreds of thousands of texts processed in total. During the dataset development process, the team conducted strict privacy protection measures and thorough ethical reviews on the data. This benchmark is primarily intended to evaluate and improve the language processing capabilities of models in the medical domain, particularly their performance when dealing with sensitive medical information.