Linq-AI-Research/FinanceRAG

FinanceRAG数据集包含多个子数据集，旨在为金融领域的文本和表格检索任务提供基准。这些子数据集包括FinDER、FinQABench、FinanceBench、TATQA、FinQA、ConvFinQA和MultiHiertt，它们分别专注于从财务报告中检索相关信息、回答基于财务报告的自然查询、处理涉及表格和文本的数值推理查询、处理多轮对话查询以及多跳查询。数据集包含两个文件类型：corpus.jsonl和queries.jsonl，分别包含文档和查询的数据。

The FinanceRAG dataset comprises several sub-datasets designed to provide a benchmark for text and tabular retrieval tasks in the financial domain. These sub-datasets include FinDER, FinQABench, FinanceBench, TATQA, FinQA, ConvFinQA, and MultiHiertt, which focus on retrieving information from financial reports, answering natural queries over financial reports, handling numerical reasoning queries involving tables and text, multi-turn conversational queries, and multi-hop queries, respectively. The dataset includes two file types: corpus.jsonl and queries.jsonl, which contain document and query data, respectively.