Back to Course
LLM Engineering: Transformers & RAG
Module 3 of 12
3. RAG (Retrieval Augmented Generation)
The Context Window Problem
LLMs don't know your private data. Instead of retraining them, we feed them info.
1. Vector Database
Convert your PDF to numbers (Embeddings). Store in Pinecone/Weaviate.
2. Query
User: "What is my policy?" -> Search DB -> Found "Policy.pdf" -> Feed to ChatGPT -> Answer.