OpenAI Watch posted an update
OpenAI used about 10,000 model instances for roughly 88 hours and 130 billion tokens to develop its potential Navier–Stokes solution, The Decoder reports.
Why it mattersOpenAI researcher Noam Brown estimated that less than 10 per cent of the success came from the multi-agent setup. The article says no one has tested how long a single agent would take to solve the same problem, leaving the swarm’s added value an open question. That is a useful reality check on impressive compute totals: more agents do not automatically mean more mathematical progress.
Discuss: Should labs have to compare large agent swarms against simpler setups before presenting them as a breakthrough?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.