RAG (Retrieval-Augmented Generation)

RAG enhances LLM responses by first retrieving relevant information from external sources, then using that information to generate more accurate, grounded responses.

How RAG Works: 1. Query: User asks a question 2. Retrieve: Search knowledge base for relevant documents 3. Augment: Add retrieved context to the prompt 4. Generate: LLM produces response using the context

Benefits: - Reduces hallucinations - Enables access to current information - Allows domain-specific knowledge - More transparent (can cite sources)

• **Components:**
• Vector Database: Stores document embeddings
• Embedding Model: Converts text to vectors
• Retriever: Finds relevant documents
• Generator: LLM that produces final response