No. 1 is terrible advice, and well illustrates what’s wrong with a lot of the deep learning crowd. It practically guarantees you’ll wind up stuck in a local optimum.

→ View original post on X — @pmddomingos