AI Foundationsintermediate12 min · 9 steps
Retrieval: Keyword, Semantic, Hybrid and HyDE
The R in RAG: how to find the right passages — by words, by meaning, by both, or by first imagining the answer.
requestdatatoken
metrics are illustrativeProblemStep 1 of 9
The model can only use what you retrieve
Retrieval is step one of RAG: given a question, fetch the k most relevant passages (top-k) from a knowledge base and put them in the prompt.
If the right passage isn't in the top-k, even the best LLM will answer from the wrong evidence. Retrieval quality is measured with recall (did we find the relevant passages?) and precision (how much of what we returned is relevant?).
InsightMost bad RAG answers are retrieval failures. Measure retrieval on its own before tuning prompts.
← Previous lessonGenetic Algorithms: Selection, Crossover, MutationNext lesson →The Augmented LLM: Retrieval, Tools and Memory
Lesson 27 of 34 in AI Foundations