NAT#

This example demonstrates how to use NVIDIA’s Context-Aware RAG (CA-RAG) system with NAT (NeMo Agent Toolkit) for document processing and question answering. The example shows how to:

  • Ingest documents using NAT

  • Perform question answering using NAT

Prerequisites#

  1. NVIDIA API Keys:

    • Get your API key from: build.nvidia.com

    • Export the following environment variable:

      export NVIDIA_API_KEY=your_api_key

  2. Install NAT and Context-Aware RAG:

    • Follow the installation instructions in the NAT Plugin Guide

    • Make sure you have both NAT and Context-Aware RAG installed in your environment

  3. Set up pre-requisites containers:

Setup#

  1. Start the NAT services:

    • Start Ingestion Service:

      nat serve --config_file=./packages/vss_ctx_rag_nat/src/vss_ctx_rag/plugins/nat/nat_config/workflow/config-ingestion-workflow.yml --port 8000

    • Start Retrieval Service:

      nat serve --config_file=./packages/vss_ctx_rag_nat/src/vss_ctx_rag/plugins/nat/nat_config/workflow/config-retrieval-workflow.yml --port 8001

Usage#

  1. Document Ingestion:

    • Documents are processed and uploaded to the NAT ingestion service

    • The system maintains document order

  2. Question Answering:

    • Send questions to the NAT retrieval service

    • Receive answers based on the ingested document content

Example Notebook#

The qna_nat.ipynb notebook provides a step-by-step walkthrough of the entire process, including:

  • Service initialization

  • Document processing and ingestion

  • Question answering examples

Notes#

  • Make sure to stop the NAT services when you’re done by pressing Ctrl+C in the terminal windows