Deep dive · · 6 min read
Most RAG failures are faithful answers to bad evidence
Six posts on reranking, citations and decomposition, and the uncomfortable thing they turn out to have in common.
Topic
Retrieval
9 entries and 1 reference.
In the glossary: Retrieval-augmented generation (RAG), Grounding, what each means and how to say it in a review
Part of Data & Databases: all Data & Databases entries · the Data & Databases lens on the map
Entries
Deep dive · · 7 min read
No model can reason about evidence your retriever never returned
Six days of retrieval notes, and not one of the failures announced itself.
Explainer · · 2 min read
Test retrieval on its own or you will tune the wrong thing
Most "the model is wrong" problems are recall problems. Hybrid search, query rewriting and versioned indexes are where they get fixed.
Explainer · · 2 min read
Why one user question may need many searches
Why one user question may need multiple searches before the evidence is good enough.
Explainer · · 2 min read
Why RAG needs both broad context and precise evidence
Why good RAG often needs both precise snippets and enough surrounding context.
Explainer · · 2 min read
Why reranking is where retrieval becomes useful
How reranking turns broad retrieval candidates into evidence the answer can depend on.
Explainer · · 2 min read
Why relevant data can still be unauthorized data
Why relevant evidence is still wrong evidence if the user was not allowed to see it.
Explainer · · 2 min read
Why keyword search and semantic search both matter
Why exact words and semantic meaning both matter in production retrieval.
Explainer · · 2 min read
Why hybrid search is often the practical default
Why combining retrieval signals is often more practical than betting on one search method.
References
- Read
RAG at Scale: Architecture and Lessons
Google Cloud · Deep dive · A production-oriented walkthrough of retrieval architecture choices and their operational cost.