Generate RAG Responses

Interacting with a Stored RAG Pipeline in GenAI Studio #

Retrieval-Augmented Generation (RAG) combines the power of a language model with external document retrieval to provide richer and more accurate responses based on a broader knowledge base.

Prerequisites #

• Access to a stored RAG pipeline connected to a data vector store.

How to Generate Responses Using a RAG Pipeline #

Navigating to GenAI from a Cluster

If you are starting from an Machine Learning Development Environment cluster, select GenAI near the lower left corner to open GenAI Studio in your browser.

• Navigate to GenAI Studio.
• Select Create a new project.
• Provide a name and then select Create. You’ll see the new project in your project list.
• Select your project from the project list to open it.
• In the Snapshots tab of your project, select New Playground to open a new snapshot Playground.

  

• In the playground, choose Document Retrieval from the playground type.

  

• In Select RAG, choose your RAG pipeline and then select Load Now.
• In the input prompt, type a question and then select Generate.
• GenAI Studio displays a response and the documents used to generate it.

Snapshot Mode

RAG snapshots are easy to identify. Once you’ve created a document retrieval snapshot, the mode column will display rag.