My RAG evals said 100%. The answers were still wrong.What a 30-question test set caught in my notes app: a reranker reading the wrong part of a note, and a phone number reaching the model.11 min read·trialswithtech.hashnode.dev33m ago
What I learned building a RAG notes app (and the evals I skipped)Embeddings, vector search, reranking, and getting Flutter, Spring Boot and Python to talk to each other13 min read·trialswithtech.hashnode.dev1ANov 6
