FlagOpen 开源 SciSage：多智能体科学综述生成与 SurveyScope 基准
GitHub - FlagOpen/SciSage: SciSage: multi-agent scientific survey generation with hierarchical reflection; includes the SurveyScope benchmark. · GitHub
智源 FlagOpen 开源 SciSage，用多智能体与分层反思生成科学综述，并附带 SurveyScope 基准。它支持从 arXiv 多源抓取论文、AI 理解论文并生成大纲、按章节生成带引用的结构化内容，兼容 GPT-4、本地模型和云服务，采用 MIT 许可证。
正文
当前显示原文，中文全文译文尚未就绪。
FlagOpen; SciSage
Repository navigation
NotificationsYou must be signed in to change notification settings; Fork4 (4); Star9 (9)You must be signed in to star a repository
SciSage: multi-agent scientific survey generation with hierarchical reflection; includes the SurveyScope benchmark.
Repository files navigation
README; MIT license
This is the official repository of SciSage.
Features
Multi-source Paper Extraction: Robust crawling from arXiv with fallback mechanisms; Intelligent Analysis: AI-powered paper understanding and outline generation; Structured Content Generation: Section-wise detailed analysis with proper citations; Multi-model Support: Compatible with GPT-4, local models, and cloud services
Edit core/model_factory.py:
Pipeline Settings
Edit core/configuration.py and set your Default model you want to use.
Afater setup server, you can test the server
Project Structure
License
MIT License - see LICENSE file for details.
Citation
If you use SciSage or SurveyScope in your research, please cite:
SciSage: multi-agent scientific survey generation with hierarchical reflection; includes the SurveyScope benchmark.