Cleanlab/FinQA-hallucination-detection

FinQA Hallucination Detection数据集是从FinQA数据集中选取的一个子集，每个样本由用户查询（金融问题）以及基于提供的上下文（原始FinQA的财务报表和表格）生成的LLM回答组成，并为每个回答标注了正确与否。该数据集适用于金融领域中检索增强生成（RAG）的虚构检测技术的基准测试。

The FinQA Hallucination Detection dataset is created from a subset of the FinQA dataset, consisting of user queries (financial questions) and LLM-generated responses based on provided context (financial statements and tables from the original FinQA), each labeled with correctness. This dataset is suitable for benchmarking hallucination detection techniques in Retrieval-Augmented-Generation (RAG) within the financial domain.