Your question is RAG Hallucination Prevention. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How do you structure a RAG pipeline to prevent model hallucinations when retrieving multi-page financial documents and tables at JPMorganChase?
Explain how you would preserve document structure, retrieve relevant evidence, generate grounded answers, and handle questions that cannot be answered from the source material. Include safety controls, evaluation methods, and practical cost and latency considerations.