Lesrning about an LLM through experimentwlly prompting it makes sense – the mistake people frequently make is asking it direct questions like "what's your context length" or what architecture do you use" and then believing what it tells them

→ View original post on X — @simonw,