Getting Deepseek R1 to already think at human levels using RAG? I will say I am currently using R1 thinking and piping through to gpt-4o structured output to see what happens. It's good, but the human made data, or rather human perfected data we have is still 3x. ---- The ultimate will be when can do RAG with reasoning. I.e. 2-3 paragraphs in, inject data elsewhere, and make it think more like a human. I am trying this thouroughly now, brb