Trying to understand modern ML systems (at the moment mostly foundation models) from a top down perspective.
Analogous to Human Psychology (Top Down) vs Human Neuroscience (Bottom Up)