耀仔AI工程垂类大模型训练数据集

本数据集源于公司自研平台耀仔AI系统，通过对智能体的交互，快速查找问题对应的规范，项目具体的管理情况，来辅助项目精细化管理。工程垂类大模型，生成报告+规范精准查询，秒级输出专业决策依据解决“判”的薄弱，作为工程领域专属智能助手,覆盖项目管理、技术指导、数据统计等场景,支持多模态交互,助力工地人员高效决策,降低工程管理操作门槛。数据集为系统的标注数据，人工对模型的问题回答进行正负例答案标注，用于对模型进行微调训练，使模型能够理解复杂工程场景、精准引用条款、具备更好的逻辑推理能力。This dataset is derived from the company's self-developed Yaozai AI system. By interacting with the built-in AI Agent, it enables rapid retrieval of relevant specifications and specific project management details to support refined project management. Equipped with an engineering-domain large language model, the system can generate reports and conduct precise specification queries, outputting professional decision-making basis within seconds, thereby addressing the shortcomings in engineering judgment. Serving as an exclusive AI assistant for the engineering field, it covers scenarios including project management, technical guidance, data statistics and more, supports multimodal interaction, helps on-site construction personnel make efficient decisions, and lowers the operational threshold for engineering management. This dataset consists of annotated data from the system: manual annotation of positive and negative answer examples for the model's responses, which is used for fine-tuning the large language model to enable it to comprehend complex engineering scenarios, accurately cite relevant clauses, and improve its logical reasoning capabilities.

搜集汇总

背景与挑战

背景概述

该数据集是由广东鼎耀工程技术有限公司基于自研耀仔AI系统生成的标注数据，专用于工程垂类大模型的微调训练。数据集包含问题、对应文档、正例答案和负例答案等字段，旨在帮助模型理解复杂工程场景、精准引用规范条款，并提升逻辑推理能力，从而辅助项目精细化管理。其应用场景覆盖建设方、施工方、设计单位等多方需求，可一键问答项目进展、生成合规技术方案与报告，有效降低工程管理操作门槛。