A 1T-parameter MoE reached 1,300 tokens per second per GPU on a public European supercomputer
๐ November 12, 2026 โข domyn.com Domyn, the Milan company leading the EUROPA consortium, published a technical post describing training trials for a mixture-of-experts model on the JUPITER supercomputer. The same write-up appeared on the PyTorch blog the same day. According to the post, these are tโฆ
Read the full story at DEV Community โ Machine Learning โ
Timeline ยท 1 report
- 2026-11-15 15:00 ยท DEV Community โ Machine Learning
A 1T-parameter MoE reached 1,300 tokens per second per GPU on a public European supercomputer