# lightrag-simple-fast-rag-hkuds-2024-10-08

## Veille

LightRAG - Simple and Fast RAG - Knowledge Graphs - Dual-Level Retrieval - EMNLP2025 - GitHub

## Titre Article

HKUDS/LightRAG: [EMNLP2025] "LightRAG: Simple and Fast Retrieval-Augmented Generation"

## Date

2024-10-08

## URL

https://github.com/HKUDS/LightRAG

## Keywords

knowledge-graph, gpt, rag, gpt-4, large-language-models, llm, genai, retrieval-augmented-generation, graphrag, dual-level retrieval, vector storage, Neo4j, PostgreSQL, Faiss, document processing, multimodal RAG, EMNLP2025

## Authors

Zirui Guo, Lianghao Xia, Yanhua Yu, Tu Ao, Chao Huang (HKUDS - Hong Kong University Data Science)

## Ton

**Profile**: Academic-technical, institutional research voice, descriptive-technical register, expert level.

**Description**: HKUDS researchers adopt the tone of an academic GitHub repository combining technical documentation with research-paper positioning (accepted at EMNLP 2025). The specialized ML/AI language (dual-level retrieval, knowledge graphs, vector storage) targets researchers and practitioners. The typical open-source project structure (README, examples, benchmarks) facilitates adoption. The technical, objective tone avoids marketing hype in favor of demonstrating capabilities. Typical of academic code publications (HuggingFace, Papers with Code style) linking research contributions to practical implementation, aimed at the ML community.

## Pense-betes

- **22k** GitHub **stars**, **3.3k forks**, MIT license
- Presentation at **EMNLP 2025** (major NLP conference)
- **Dual-level retrieval**: combination of knowledge graphs + vector databases
- **Multimodal** support via RAG-Anything (text, images, tables, equations)
- **25+ configuration parameters** for fine-grained customization
- LLM recommendations: **minimum 32B parameters**, context **32KB+** (64KB ideal)
- **6 query modes**: local, global, hybrid, naive, mix, bypass
- **Multiple storage** support: PostgreSQL, Neo4j, Redis, MongoDB, Milvus, Qdrant, Faiss
- **RAG-Anything** integration for document processing (PDF, PPT, CSV)
- **Advanced management**: deletion by document ID, by entity name
- **Streaming responses** and conversation history
- **Citation** support for source attribution
- **Docker Compose** available for quick deployment
- Active community: Discord, WeChat, comprehensive documentation

## RésuméDe400mots

LightRAG is a Retrieval-Augmented Generation (RAG) system developed by HKUDS (Hong Kong University Data Science) that will be presented at EMNLP 2025, a major conference in natural language processing. With **22k** GitHub **stars** and under an MIT license, the project aims to simplify and accelerate knowledge extraction and intelligent querying through an innovative approach combining knowledge graphs and vector databases.

**Dual-Level Retrieval Architecture**

LightRAG's core innovation lies in its "dual-level retrieval" system, which merges two complementary approaches: knowledge graphs to capture structural relationships between entities, and vector databases for semantic search. This hybridization offers **six distinct query modes** (local, global, hybrid, naive, mix, bypass) that adapt the retrieval strategy to context and specific needs.

**Multimodal Support and Multiple Formats**

The recent integration with RAG-Anything considerably broadens document-processing capabilities. The system now seamlessly handles text, images, tables, and equations from various formats (PDF, DOC, PPT, CSV). This multimodal approach turns LightRAG into a complete solution for knowledge extraction from heterogeneous enterprise documents.

**Storage Flexibility and Scalability**

LightRAG supports a modular four-tier storage architecture: KV storage for LLM caching and text chunks (JSON, PostgreSQL, Redis, MongoDB), Vector storage for embeddings (NanoVectorDB, Milvus, Chroma, Faiss, Qdrant, PostgreSQL, MongoDB), Graph storage for the entity-relationship graph (NetworkX, Neo4J, PostgreSQL, AGE), and Document status storage for processing tracking. This flexibility allows choosing the optimal backend based on performance and infrastructure requirements. For high-performance production, Neo4J is recommended as the graph database.

**LLM Requirements and Recommendations**

The recommended specifications reflect the system's sophistication: an LLM with **a minimum of 32 billion parameters** and a context length of **32KB** (64KB preferred) to optimize entity-relationship extraction. Mainstream multilingual embedding models such as BAAI/bge-m3 and text-embedding-3-large are suggested. The system supports OpenAI (GPT-4o, GPT-4o-mini), Ollama for local models, and Hugging Face.

**Advanced Configuration and Customization**

LightRAG exposes **more than 25 initialization parameters** offering granular control: chunk size (default 1200 tokens), entity-extraction loops (default 1), embedding batch size (default 32), LLM cache parameters, and token budgets for entities, relationships, and total context. This configurability allows fine-tuning the performance/cost trade-off.

**Advanced Features and Management**

The system offers sophisticated capabilities: document deletion by ID, entity deletion by name (maintaining graph/vector consistency), citation support for source attribution, streaming responses, conversational history tracking, and insertion of custom knowledge graphs. The asynchronous architecture (asyncio) ensures performance and scalability.

**Deployment and Community**

Flexible installation via PyPI (`pip install "lightrag-hku[api]"`), from source, or via Docker Compose with pre-configured environment files. The active community maintains Discord, WeChat, comprehensive documentation, examples, and tutorial videos. Recent updates (PostgreSQL support, citations, deletions, Neo4J) demonstrate active development.

## GrapheDeConnaissance

- HKUDS —a_créé→ LightRAG (TECHNOLOGIE, 0.99)
- LightRAG —observé_dans→ EMNLP2025 (EVENEMENT, 0.98)
- LightRAG —utilise→ dual-level retrieval (CONCEPT, 0.99)
- dual-level retrieval —utilise→ knowledge graphs (TECHNOLOGIE, 0.97)
- dual-level retrieval —utilise→ bases de données vectorielles (TECHNOLOGIE, 0.97)
- LightRAG —utilise→ RAG-Anything (TECHNOLOGIE, 0.95)
- RAG-Anything —permet→ traitement multimodal (CONCEPT, 0.93)
- LightRAG —utilise→ Neo4j (TECHNOLOGIE, 0.96)
- LightRAG —utilise→ PostgreSQL (TECHNOLOGIE, 0.96)
- LightRAG —recommande→ modèles 32B paramètres minimum (CONCEPT, 0.92)
- LightRAG —permet→ 6 modes de requête (CONCEPT, 0.95)
- HKUDS —utilise→ GitHub (TECHNOLOGIE, 0.97)
- LightRAG —mesure→ 22k étoiles GitHub (MESURE, 0.98)

---
Canonical: https://www.thekb.eu/en/fiches/lightrag-simple-fast-rag-hkuds-2024-10-08/