6. Studying emergent behavior in language models. – How/why does scaling unlock emergence?
– Can we predict future emergent capabilities?
– How do we get these abilities in smaller models?

→ View original post on X — @_jasonwei,