RAG vs. Long Context Models: Is Retrieval-Augmented Generation Dead?

[@enmukeeenmukee3254]

Today, relatively small organizations deal with TBs of data. I cannot understand how RAG is nothing but a toy if 128K or even 1M token is considered large.

[@WhatsAI]

You don’t need to send all those TB for each query to the LLM. If you do, then you need fine tuning, not rag or long context!