From the course: Azure for Developers: Retrieval-Augmented Generation (RAG) with Azure AI
Join today to access over 24,600 courses taught by industry experts.
RAG process and architecture - Azure AI Services Tutorial
From the course: Azure for Developers: Retrieval-Augmented Generation (RAG) with Azure AI
RAG process and architecture
- [Instructor] Now that we understand the basic concept of RAG and the concepts of token and embeddings, let us discuss how RAG works. The first process is normally chunking, which allows large documents to be broken down into smaller documents. This process is important so that information is not lost. Imagine a document 1,000 pages long that is summarized into just two pages. Important details and information would be lost. If a page has a lot of information, you may need to break down the page into smaller pages to ensure details are not lost. It is normally the case that after chunking, you may end up with more pages than what you originally started. Two, the chunk documents are then converted into embeddings. We'll use the Azure OpenAI embeddings model in this course. The embeddings are then stored in a data source for storage and retrieval. This data source will serve as a systems knowledge base. Steps one and two are primarily done during the initial setup or at a specified…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.