RUC-NLPIR/FlashRAG_datasets

FlashRAG数据集是一个用于检索增强生成（RAG）研究的Python工具包，包含了32个预处理的基准RAG数据集和13种最先进的RAG算法。这些数据集涵盖了多种任务类别，如问答、摘要生成、文本生成等，并且所有数据集都以统一的格式进行了预处理，以便于使用。每个数据集都包含了训练集、开发集和测试集，具体大小在README中有详细列出。此外，还提供了用于检索的文档库的路径和格式说明。

The FlashRAG dataset is a Python toolkit for Retrieval-Augmented Generation (RAG) research, containing 32 pre-processed benchmark RAG datasets and 13 state-of-the-art RAG algorithms. These datasets cover various task categories such as question answering, summarization, and text generation, and all datasets have been pre-processed to ensure a consistent format for ease of use. Each dataset includes training, development, and test sets, with specific sizes detailed in the README. Additionally, the toolkit provides a document corpus for retrieval, with path and format instructions included.

原始信息汇总

FlashRAG Datasets

基本信息

• 许可证: CC BY-SA 4.0
• 任务类别:
  • 问答
  • 摘要生成
  • 文本到文本生成
• 语言: 英语
• 数据集大小: 1M < n < 10M

数据集名称

• 名称: FlashRAG Datasets