Railway Training Dataset (RTD)
收藏资源简介:
Railway Training Dataset (RTD)是由北京交通大学自动化与智能学院构建的铁路领域特定微调数据集,包含10100条结构化问答对,涵盖铁路法律条款、铁路规章和铁路专业知识。数据集基于中国铁路机车和车辆驾驶员资格考试大纲构建,原始文本数据包含776,000个中文Token,分为法律条款、铁路规章和铁路专业知识三大类。该数据集旨在提升大语言模型在铁路领域的专业知识,特别是在故障处理场景中的应用。通过自动化处理策略生成问答对,并经过严格的数据过滤,最终用于铁路驾驶场景的对话支持。
Railway Training Dataset (RTD) is a railway domain-specific fine-tuning dataset constructed by the School of Automation and Intelligence, Beijing Jiaotong University. It contains 10,100 structured question-answer pairs covering railway legal provisions, railway regulations and railway professional knowledge. The dataset is built based on the qualification examination syllabus for Chinese railway locomotive and vehicle drivers. The original textual data includes 776,000 Chinese Tokens, and is divided into three categories: legal provisions, railway regulations and railway professional knowledge. This dataset aims to enhance the domain-specific knowledge of Large Language Models (LLMs) in the railway field, especially their application in fault handling scenarios. Question-answer pairs were generated via automated processing strategies, and after strict data filtering, the dataset is ultimately used for dialogue support in railway driving scenarios.

- 1A Driver Advisory System Based on Large Language Model for High-speed Train北京交通大学自动化与智能学院 · 2025年



