DeepSieve offers two RAG modes, the main differences are as follows:
- Naive mode: It is suitable for simple query tasks and is fast but less accurate. It directly retrieves and fuses answers to sub-questions and is suitable for scenarios that do not require complex reasoning.
- Graph mode: It is suitable for complex queries and multi-step reasoning with higher accuracy but increased computational cost. It handles dependencies between sub-problems through graph structures and is suitable for scenarios that require deep analysis.
Users can use the environment variable RAG_TYPE Switching modes (naive maybe graph), and choose according to the needs. For example, multidimensional problems in business data analysis may be better suited to Graph mode, while simple data queries can use Naive mode.
This answer comes from the articleDeepSieve: a RAG Intelligent Information Screening Tool for Processing Complex Query SourcesThe
May not be reproduced without permission:AI productivity tools " What is the difference between Naive mode and Graph mode in DeepSieve?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil