Not RAG: asking the LLM to generate using its internal weights only

RAG: providing the LLM with contextual data you’ve pulled from outside its weights that you believe relate to a query