The reality however is that unless you're spending tens of thousands on hardware or setting up cloud instances you can't run anything near close to state of the art.
In almost all cases right now, a subscription with a cloud AI wins.
> In almost all cases right now,
What sort of benchmark or test would convince you that "almost all" is more like "your weird set of cases" and nowhere near "almost all"?
> In almost all cases right now,
What sort of benchmark or test would convince you that "almost all" is more like "your weird set of cases" and nowhere near "almost all"?