Pipeline 3: Knowledge graph index-based RAG

It’s time to create a knowledge graph index-based RAG pipeline and interact with it. As illustrated in the following figure, we have a lot of work to do:

Figure 7.5: Building knowledge graph-index RAG from scratch

In this section, we will:

• Generate the knowledge graph index
• Display the graph
• Define the user prompt
• Define the hyperparameters of LlamaIndex’s in-built LLM model
• Install the similarity score packages
• Define the similarity score functions
• Run a sample similarity comparison between the similarity functions
• Re-rank the output vectors of an LLM response
• Run evaluation samples and apply metrics and human feedback scores
• Run metric calculations and display them

Let’s go through these steps and begin by generating the knowledge graph index.

Generating the knowledge graph index

We will create a knowledge graph index from...