This post discusses the importance of evaluating retrieval in Retrieval-Augmented Generation (RAG) systems. It explores popular academic benchmarks, the effects of data blending on benchmarks, and the recommended metrics for evaluating retrievers.
Table of contents
Effects of data blending on benchmarksWhat are the popular benchmarks?How to select the best model for QA benchmarksEvaluation metricsRetrieval benchmarkConclusionSort: