The goal here is to squeeze as much power out of as small a model as possible. The smaller the model, the easier it is to update and tune—as well as run it cheaply (or even locally on a device).

→ View original post on X — @mattlynley,