This seems like it is a big step in the right direction for continual learning. I am beginning to understand the big talk out of Dario and Sholto of late. Maybe continual learning is solved?