Throwing something out to the community for a bit of an insight. I got thinking about the consumption of tokens when working with various databases and here is my understanding: When I ask as question that is essentially converted to tokens. The LLM then "reads" that and generates the response which
