Build a RAG-based QnA application using Llama3 models from SageMaker JumpStart

This article provides a step-by-step guide for creating a Retrieval-Augmented Generation (RAG) question-answering application using the Llama3-8B model from SageMaker JumpStart and the BGE Large EN v1.5 embedding model. It covers the following key topics:

Specifically, the article covers:

• An overview of Llama3-8B, BGE Large EN v1.5, and RAG
• Prerequisites and setup for the solution
• Deploying the Llama3-8B and BGE Large EN v1.5 models on SageMaker JumpStart
• Setting up the models with LangChain
• Preparing data (Amazon's Annual Reports) and generating embeddings
• Implementing different approaches for document retrieval and question answering using LangChain:
  • Regular Retrieval Chain
  • Parent Document Retriever Chain
• Cleaning up resources to avoid incurring unnecessary costs