Easy RAG with LangChain4J and Docker
summary
Demonstrates a runnable RAG pipeline that ingests a Microsoft annual report, splits it into 1,563 text segments, embeds them, and stores text+vectors in a local vector DB, all orchestrated with Easy RAG on LangChain4J and Docker. Uses local embedding models (medm / tiny Llama) or cloud text-embedding-3-small and a Qdrant-like store, retrieves top‑3 segments for prompts, and runs tests with Testcontainers; demo works but small models sometimes return incorrect answers.
AI-written summary · about
chapters (8)
- 0:00Introduction to LangChain4J
- 1:50What is LangChain4J
- 4:02Why use RAG pattern
- 10:46Data ingestion process
- 16:34Best practices for RAG
- 21:11Querying with RAG
- 26:55Optimizing model performance
- 30:04Summary and resources
related talks
48mCrafting your own RAG system: Leveraging 30+ LLMs for enhanced performance
48mCrafting your own RAG system: Leveraging 30+ LLMs for enhanced performance
25mSet Up an Open Source RAG in 30 Minutes
27mBuilding a Reproducible RAG Pipeline
59mRAG Against the Machine
34mImplementing an open-source RAG with OpenSearch
35mThe Levels of RAG 🦜
42mTo RAG or not to RAG