GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
1–10 of 467 posts
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#2Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#3Announcement: https://x.com/__eknight__/status/2075643450196971805 Prompt: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98...
> Spend at least 8 hours on this before even thinking of returning or giving up.
Do current model harnesses have concepts of amount of time spent? Sometimes the model notices if a subprocess takes too long/hangs and kills it, but I've never seen it time itself.
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#4Announcement: https://x.com/__eknight__/status/2075643450196971805 Prompt: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98...
> Spend at least 8 hours on this before even thinking of returning or giving up. Do current model harnesses have concepts of amount of time spent? Sometimes the model notices if a subprocess takes too long/hangs and kills it, but I've never seen it time itself.
Temporal awareness with GPT-Live
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#5Announcement: https://x.com/__eknight__/status/2075643450196971805 Prompt: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98...
> Spend at least 8 hours on this before even thinking of returning or giving up. Do current model harnesses have concepts of amount of time spent? Sometimes the model notices if a subprocess takes too long/hangs and kills it, but I've never seen it time itself.
No, however, if they have the ability to get the current time, they obey constraints like these in a way a model a year ago didn't.
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#6Announcement: https://x.com/__eknight__/status/2075643450196971805 Prompt: https://cdn.openai.com/pdf/04d1d1e4-bc75-476a-97cf-49055cd98...
> Spend at least 8 hours on this before even thinking of returning or giving up. Do current model harnesses have concepts of amount of time spent? Sometimes the model notices if a subprocess takes too long/hangs and kills it, but I've never seen it time itself.
The voice models certainly can't: https://kittygr.am/reel/DWr31A1B1Ux/
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#7Earlier quoted context omitted.
> Spend at least 8 hours on this before even thinking of returning or giving up. Do current model harnesses have concepts of amount of time spent? Sometimes the model notices if a subprocess takes too long/hangs and kills it, but I've never seen it time itself.
Temporal awareness with GPT-Live https://www.youtube.com/watch?v=8vvWTz6N7Qg
Fascinating! This is relative time in a continuously processing voice model, here, they're using an LLM with absolute time.
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#8"Assume for purposes of this task that a complete affirmative proof exists"
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#9It's really neat that the prompt was released!
I'm curious how many unsolved problems are tried against frontier models when they come out. Are we trying every problems against every release? What is the solve success rate? Is there a sub-community within Mathematics that is coordinating this effort? How much untapped opportunity is there here?
Re: GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]
#10But is the proof accepted to be correct? That is what distinguishes this from being notable compared to any other AI slop proof.