usail-hkust/ScIRGen-Geo

ScIRGen-Geo 数据集是一个大规模、面向任务的科学研究成果增强生成（RAG）数据集，专注于地球科学领域。该语料库完全支持双语（英文<->中文），提供两种语言的平行内容。数据集旨在反映现实世界的研究查询，包括真实的问题、详细的元数据信息和相关的论文摘要。

The ScIRGen-Geo Dataset is a large-scale, task-oriented dataset designed for retrieval-augmented generation (RAG) in scientific research, focusing on the geoscience domain. The corpus is fully bilingual (English ↔ Chinese), offering parallel content in both languages. The dataset is crafted to reflect real-world research inquiries, incorporating realistic questions, detailed dataset metadata, and relevant paper excerpts.