Report
Meta's gigawatt-scale Prometheus supercluster in Ohio will have roughly 1 million GPUs at full scale
Techmeme, citing Engineering at Meta, says Prometheus was used to train Meta's Muse foundational models.
TLDR
Techmeme shares an Engineering at Meta account of Prometheus, a gigawatt-scale supercluster in Ohio. According to Techmeme's summary, Prometheus will have roughly 1 million GPUs at full scale and was used to train Meta's Muse foundational models.
Combined views
2.2K
1 Source, first seen ago
