Improving RAG performance — A Structured Approach (Part 6(B) of RAG Series)
A comprehensive and structured approach
This is part 6 of the “Retrieval-Augmented Generation (RAG) — Basics to Advanced Series”. Links to other blogs in the series are at the bottom of this blog. Taking forward from part 1 (RAG Basics), part 2 (Chunking), part 3 (Embedding) and part 4 (Vector Databases and Vector Libraries) and part 5 (Evaluation of RAG). In this blog, we will focus on one of the most challenging but also the most important step i.e. “Improving RAG performance”
This part 6 is further divided into Part A and Part B. Part A of the blog covered the Ingestion Stage and covers key topics like Data clean up, Data Enrichment, Chunking, Embedding, Vector database and Indexing and Query enhancement and Prompt Engineering
This blog would cover the retrieval and generation stage which would include experimentation around areas like Retrieval, Re-ranking, Fine Tuning and bringing everything together.
Diagrammatically, going back to our RAG Architecture, Part A covers the components in the Green boxes and Part B covers the components in the Blue boxes.
