Retrieval-Augmented Generation

← Back to Large Language Models

Enhance LLM responses by retrieving relevant documents and injecting them into the context. Reduces hallucination, enables access to private/recent data, and provides attributable answers.

Key Properties

• Vector Databases
• Embedding Models
• Chunking Strategies
• Retrieval Methods

Architecture

1. User query → embed query
2. Search vector database for similar documents
3. Inject retrieved documents into LLM prompt
4. LLM generates answer grounded in retrieved context

Related

• Embeddings (encode documents and queries)
• Foundation Models (generate answers from context)

---

nlp llm rag retrieval