Article Source

• Overview of RAG Approaches with Vector Databases

Overview of RAG Approaches with Vector Databases

Abstract

Retrieval Augmented Generation (RAG) is a fast-moving practice to extend the knowledge of your LLMs to private and real-time data. In the past few weeks and months, different approaches have been constructed by developers around the world, including:

• Naive RAG
• Agent Assisted RAG
• Guardrails RAG
• Knowledge Graph RAG
• Time Based RAG

These often use LangChain or LlamaIndex with a vector database to create useful conversational apps.

Join this session as we discuss different approaches to RAG, as well as optimizing RAG with structured retrieval, chunk decoupling, and more.