Part 2 Creating RAG systems

 

Now that you are familiar with the fundamental idea of RAG and the components of a RAG system, the second part of the book will guide you through building a basic RAG system with the core pipelines and their evaluation. This part of the book not only offers theoretical details, but also simple code snippets that will provide you with hands-on experience in building a RAG pipeline.

In chapter 3, you’ll learn the details of the indexing pipeline and its four components: loading, chunking, embeddings, and vector storage. Each of these components has a variety of techniques to choose from. This chapter also discusses the suitability of these options for different use cases. Step by step, you’ll build an indexing pipeline and create the knowledge base for your RAG system.

Chapter 4 talks about retrievers, prompting techniques, and using LLMs for output generation. These elements form the three components of the generation pipeline: retrieval, augmentation, and generation. In this chapter, you will build the generation pipeline that interacts with the knowledge base, created using the indexing pipeline in chapter 3.