2 Comments
User's avatar
Petar Dimov's avatar

The biggest takeaway for me is that scaling only works when the architecture stays efficient enough to handle real production workloads

Ludovico Bessi's avatar

Oh yess, this is what ML@Scale is all about :)