generative-ai video for a Generative AI Engineer is comparing two different RAG configurations for a legal assistant: Config A : Uses sentence-level chunking
A Generative AI Engineer is comparing two different RAG configurations for a legal assistant: Config A : Uses sentence-level chunking and cosine similarity Config B : Uses paragraph-level chunking with semantic re-ranking The team has curated a test set of 100 legal queries with expected ideal responses. They want to select the best configuration for deployment based on objective evaluation. Which evaluation approach should the engineer use to determine the better configuration?