Understanding the steps needed to create a RAG system
To implement RAG, a sequence of steps needs to be carried out. First, we need to define the sources from which data will be retrieved. This could range from online databases, specific websites, or even customized data repositories. Then, we need to optimize and store this information for retrieval. Once we’ve done this, we can use a retrieval system to fetch the relevant information based on the query context. This information is then passed on to the language model during the prompting process. The following figure outlines the different steps:
Figure 7.1 – Steps for creating a RAG system
I appreciate that a lot is going on here, so let’s look at each step in more detail.
Defining your RAG data sources
For our first step, we must identify and select the sources from which our knowledge data will be retrieved. This could include online databases, specific websites, or customized...
