Comment by canadiantim

I'm sure this is a stupid question, but why is it much better to be caching LLM functions during development?