I think compute optimal doesn't mean performance won't continue to increase if we continue to train a model even after the compute optimal point.

→ View original post on X — @yitayml