Article Source
Evaluating RAG Performance with Vector Databases
Abstract
With the rapid evolution of RAG with vector databases and LLMs, it leaves many wondering: how do I evaluate performance? Vector databases surface similar vector embeddings from a query vector, but it can be difficult to quantify if these similar vectors are meeting your goals, especially with RAG.
In this session, we’ll overview different state-of-the-art approaches to evaluating RAG including BLEU, ROGUE, RAGAS, and more. Join for an introductory discussion on how the approaches differ and the tradeoffs in each.