Add a baseline NUM solver and a random agent to compare their performance against the trained models.
Activity