Meta says Prometheus delivered 85% goodput training Muse Wonder

Meta said its Prometheus supercluster in New Albany, Ohio, delivered more than 85% goodput during pretraining of its Muse Wonder model and more than 90% during reinforcement learning, according to an October 19 post from AI at Meta on X. Meta said firmware fixes have since raised the cluster's interruption interval beyond 54,000 GPU-hours, as the company builds toward a roughly 1 million-GPU cluster.

Meta says Prometheus delivered 85% goodput training Muse Wonder The Ohio cluster's first phase trained Muse Wonder; Meta says firmware fixes have since raised its interruption interval beyond 54,000 GPU-hours. By Ryan Merket https://runtimewire.com/author/ryan-merket · Published Primary source: X https://x.com/AIatMeta/status/2112245968750145561 Why it matters The self-reported figures put reliability alongside GPU scale in Meta's frontier-model strategy and show the operational bottleneck it is trying to address as it builds toward a roughly 1 million-GPU cluster. Meta says its Prometheus supercluster in New Albany, Ohio, delivered more than 85% goodput during Muse Wonder https://runtimewire.com/models/meta/muse-wonder pretraining and more than 90% during reinforcement learning https://runtimewire.com/models/huggingface/lobonexequiel-reinforcement-learning-daff1a058d32e2ca . AI at Meta said in an October 19th post on X https://x.com/AIatMeta/status/2112245968750145561 that the first phase of Prometheus trained Muse models, including Muse Wonder.… When you purchase through links in our articles, we may earn a small commission https://runtimewire.com/affiliate-monetization-standards . This doesn’t affect our editorial independence.