upvote
> The foundation so far has been one of brute force scaling.

In cloud AI, sure. But in the smaller open weights model territory it feels like we're well into optimisation?

reply