GH Repository · HKUDS
LightRAG
[EMNLP2025] LightRAG: Simple and Fast Retrieval-Augmented Generation
- stars
- 39,660
- 30-day movement
- +10836/day
- Related entries
- 60
- Connections
- 4
knowledge-graphpythonllmragasgraphragretrieval-augmented-generationpython/uvlarge-language-modelsgenaimakegptdockerdoclingragmineruPython
LightRAG is a Python library for simple and fast retrieval-augmented generation, published as an EMNLP 2025 paper artifact. Its topics indicate it builds knowledge graphs for GraphRAG-style retrieval, with tooling hooks for document parsers like Docling and MinerU and evaluation via Ragas.
You want a research-backed, MIT-licensed RAG framework that combines knowledge-graph retrieval with LLMs without heavy setup.
Use it to
- Build knowledge-graph-backed RAG pipelines
- Run graph-based retrieval over document collections
- Evaluate RAG output quality with Ragas
- Parse documents into retrievable form with Docling or MinerU
- Experiment with GraphRAG approaches in Python
For Python developers building LLM retrieval pipelines
- Role
- rag
- Language
- Python
- Licence
- MIT
- Forks
- 5,584
- Open issues
- 189
- Last push
- 2026-09-15
- Latest release
- v1.0.3 · 2024-12-05
topicsragknowledge-graphgraphragllmretrieval-augmented-generationgenai