Best Practices#

This section provides best practices for configuring the Tokkio LLM-RAG agent and the corresponding RAG server.

RAG Server#

The Tokkio LLM-RAG agent supports any external RAG server that adheres to the schema from NVIDIA RAG Blueprint. The RAG server can be tuned for latency or accuracy, depending on the use case. Best practices for common NVIDIA RAG Blueprint settings can be found here: NVIDIA RAG Blueprint Best Practices.