Slightly odd choice of models to benchmark, isn't it?

→ View original post on X — @petergostev