zlacker

[return to "Pushing ChatGPT's Structured Data Support to Its Limits"]
1. engina+og[view] [source] 2023-12-27 16:41:40
>>goranm+(OP)
is the tipping thing correct? I provided the same prompt to ChatGPT and received multiple emojis without offering a tip.

prompt: you're Ronald McDonald. respond with emojis. what do you do for fun? answer::circus_tent::hamburger::juggling::party_popper::balloon::game_die::french_fries::performing_arts::rolling_on_the_floor_laughing::people_holding_hands::rainbow::art_palette:

◧◩
2. minima+ii[view] [source] 2023-12-27 16:50:46
>>engina+og
Your mileage may vary with any examples since ChatGPT at a nonzero temperature is nondeterministic.

If that example is through the ChatGPT web UI and not the ChatGPT API then that's a different story entirely.

◧◩◪
3. engina+xj[view] [source] 2023-12-27 16:57:49
>>minima+ii
yes, I've used ChatGPT. API allows temperature to be configured. Is there a reason to offer tips?
◧◩◪◨
4. minima+jk[view] [source] 2023-12-27 17:02:41
>>engina+xj
The point is you do not have a valid counterexample since you are using a different workflow than what's described in the article.

In my personal experience working with more complex prompts with more specific constraints/rules, adding the incentive in the system prompt has got it to behave much better. I am not cargo-culting: it's all qualitative in the end.

[go to top]