Meta says Prometheus delivered 85% goodput training Muse Wonder

Meta says its Prometheus supercluster in New Albany, Ohio, delivered more than 85% goodput during Muse Wonder pretraining and more than 90% during reinforcement learning. AI at Meta said in an October 19th post on X that the first phase of Prometheus trained Muse models, including Muse Wonder. Meta's engineering account adds that the cluster recorded about 47,000 GPU-hours between interruptions during those training runs.