gusye1234/nano-graphrag
A minimal, hackable GraphRAG implementation for knowledge graph-based retrieval-augmented generation.
Not currently ranked — collecting fresh signals.
star history
This is a simplified, educational implementation of Microsoft’s GraphRAG system. It builds knowledge graphs from documents to enable structured retrieval over unstructured text. The system leverages LLMs to extract graph entities and relationships, then combines graph traversal with vector similarity search for improved context. It supports async operations and integrates with FAISS, Neo4j, and Ollama for vector and graph storage backends.
Frequently asked
- What is gusye1234/nano-graphrag?
- A minimal, hackable GraphRAG implementation for knowledge graph-based retrieval-augmented generation.
- Is nano-graphrag open source?
- Yes — gusye1234/nano-graphrag is open source, released under the MIT license.
- What language is nano-graphrag written in?
- gusye1234/nano-graphrag is primarily written in Python.
- How popular is nano-graphrag?
- gusye1234/nano-graphrag has 3.9k stars on GitHub.
- Where can I find nano-graphrag?
- gusye1234/nano-graphrag is on GitHub at https://github.com/gusye1234/nano-graphrag.