Gradient-based learning causes distributed representations to emerge.

→ View original post on X — @ylecun