The modular design of RAGLight allows users to choose different working modes according to their needs: the standard RAG mode implements the basic retrieval-generation process; the Agentic RAG adds the Intelligent Body feature, which supports multi-step reasoning and dynamic retrieval strategy adjustment; and the RAT (Retrieval Augmented Thinking) mode further enhances the logic and accuracy of the answers through the introduction of a reflection step. These three modes are realized through different pipeline classes (e.g., RAGPipeline, AgenticRAGPipeline, RATPipeline), providing users with a full set of RAG solutions from basic to advanced.
This answer comes from the articleRAGLight: Lightweight Retrieval Augmentation Generation Python LibraryThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil