This is a special post for quick takes by MarkelKori. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Comment Permalink
Navier-Stokes is solved by OpenAI: https://openai.com/index/navier-stokes-solution/ Around 10,000 (!) agents worked on it simultaneously.
Showing 3 of 7 replies (Click to show all)
To build on @Garrett Baker comment, there's public evidence that it's an Astra fork: https://xcancel.com/chi_t_williams/status/2097440947889954974#m.
OpenAI's internal model that solved Navier Stokes was probably in training before August 28. That day, the company restarted its "large frontier RL run" that it had previously paused while dealing with the hugging face incident.
... (read more)OpenAI claimed today that the Navier Stokes solution came from a swarm of AIs of a new internal model. They write "Since August 28 we have been training a new internal model" that's a l
2roha
https://cims.nyu.edu/~tristanb/statement.pdf
[...]
[...]
https://openai.com/index/navier-stokes-solution
[...]
The OpenAI agents may have found the solution along the specific research direction of Tristan Buckmaster and Levent Alpöge because their model may have been trained on their private data containing all their drafts.
4Tapatakt
If that's a case it's a wonderful way to work around pause commitments! You promise "a two-week pause in reinforcement learning (RL) training on our latest models intended for deployment", then you fork your model, say that this fork isn't intended for deployment, and continue training. And hey, in the future you can change your mind and deploy it after all!