Better RAG 3: The text is your friend

olickel.com · 1,538 words · saved by 2 readers

This is part 3 of a series on Retrieval Augmented Research, or how to build better RAG systems to answer complex questions. Part 1 covers the basics and key concerns when building RAG systems, and Part 2 covers multi-turn retrieval, or 'walking'. In this part, we'll go over transformations that can be applied to your source data and the question, to improve retrieval. A lot of RAG pipelines embody the paths their creators took to discovering embeddings - including mine - and a lot of us fall into the same trap. First, you find embeddings and leave the world of fuzzy search behind. Maybe you've heard of BM25, or perhaps Elastic or Postgres were the closest you had - but embeddings are great! True semantic search for the first time! Much like any addictive new tool you start to use and then overuse it. Before you know it you've embedded everything you can find, and you're retrieving too much. It works sometimes, and other times nothing works. Everything you search for has thousands of ma

Better RAG 3: The text is your friend Hrishi Olickel About me Better RAG 3: The text is your friend Use the whole brain, luke This is part 3 of a series on Retrieval Augmented Research, or how to build better RAG systems to answer complex questions. Part 1 covers the basics and key concerns when building RAG systems, and Part 2 covers multi-turn retrieval, or 'walking'. In this part, we'll go over transformations that can be applied to your source data and the question, to improve retrieval. A lot of RAG pipelines embody the paths their creators took to discovering embeddings - including mine

Better RAG 3: The text is your friend

saved by

related reading