3 ms·
I use GPT-4 pretty consistently(set up a discord bot for myself). What I found myself doing was tending towards the most simple prompt that the LLM would still
by alpark3 3y ago
I use GPT-4 pretty consistently(set up a discord bot for myself). What I found myself doing was tending towards the most simple prompt that the LLM would still understand - If I asked a human expert the types of prompts I was giving GPT, I most likely would've gotten a clarifying question rather than an answer like the LLM was giving me, simply because I'm talking in such short and concise sentences.
I think the interesting thing is that the more concise a message is to a fellow human, the more work needs to be done by the other party in order to actually decode my message, even if it is ultimately understandable. Whereas with LLMs, shorter token length doesn't really matter: matrices of the same size are being multiplied anyways.
- LegitShady 3y agoI think because a human actually wants to figure out what you want, and a you're just going to keep prompting that ML model until you get something similar to what you want, something that would annoy a human and probably waste their time or make an endeavor extremely expensive. I don't think its really fundamental to LLMs its just that you don't treat a human the same way you treat an unthinking unfeeling computer system whose transactions are cheap and relatively near instant compared to requesting from a human.
- saxelsen 3y agoThis touches on what I think is a main separator between the GPT models and humans: If a human is unsure of the instructions, they will ask clarifying questions. GPT-4 cannot do that currently.
- travisjungroth 3y agoIt can. I’ve had it ask questions without explicitly being told to do so. Combine Chain of Thought and ReAct with question being an action and it will happen regularly.