News for hackers. Stuff that matters.
Re:OpenAI o1 system card by jsheard on Thursday December 05, 2024 @01:23PM (#42330894)
Do they still threaten to terminate your account if they think you're trying to introspect its hidden chain-of-thought process?
Re:OpenAI o1 system card by visarga on Thursday December 05, 2024 @01:56PM (#42331293)
A few days ago the QwQ-32B model was released, it uses the same kind of reasoning style. So I took one sample and reverse engineered the prompt with Sonnet 3.5. Now I can just paste this prompt into any LLM. It's all about expressing doubt, double checking and backtracking on itself. I am kind of fond of this response style, it seems more genuine and openended.

https://pastebin.com/raw/5AVRZsJg

Re:OpenAI o1 system card by thegabriele on Thursday December 05, 2024 @04:10PM (#42332778)
I tried this with LeChat (mistral) and ChatGPT 3.5 (free) and they start to respond to "something" following the style but... without any question asked.
Re:OpenAI o1 system card by AlfredBarnes on Thursday December 05, 2024 @03:52PM (#42332618)
Thank you for doing that work, and even more for sharing it. I will have to try this out.
Re:OpenAI o1 system card by marviel on Thursday December 05, 2024 @03:39PM (#42332496)
Thanks, I love this
Re:OpenAI o1 system card by rsync on Thursday December 05, 2024 @04:23PM (#42332921)
An aside ...

Isn't it wonderful that, after all of these years, the pastebin "primitive" is still available and usable ...

One could have needed pastebin, used it, then spent a decade not needing it, then returned for an identical repeat use.

The longevity alone is of tremendous value.

Re:OpenAI o1 system card by SirYandi on Thursday December 05, 2024 @04:29PM (#42332975)
And then once the answer is found an additional prompt is given to tidy up and present the solution clearly?
Re:OpenAI o1 system card by RestartKernel on Thursday December 05, 2024 @04:40PM (#42333100)
Interestingly, this prompt breaks o1-mini and o1-preview for me, while 4o works as expected — they immediately jump from "thinking" to "finished thinking" without outputting anything (including thinking steps).

Maybe it breaks some specific syntax required by the original system prompt? Though you'd think OpenAI would know to prevent this with their function calling API and all, so it might just be triggering some anti-abuse mechanism without going so far as to give a warning.

Re:OpenAI o1 system card by int_19h on Thursday December 05, 2024 @07:19PM (#42334453)
A prompt is not a substitute for a model that is specifically fine-tuned to do CoT with backtracking etc.
Re:OpenAI o1 system card by foundry27 on Thursday December 05, 2024 @04:20PM (#42332891)
Weirdly enough, a few minutes ago I was using o1 via ChatGPT and it started consistently repeating its complete chain of thought back to me for every question I asked, with a 1-1 mapping to the little “thought process” summaries ChatGPT provides for o1’s answers. My system prompt does say something to the effect of “explain your reasoning”, but my understanding was that the model was trained to never output those details even when requested.