IRL 7/8: Generalizing human-robot cooperation: Cooperative IRL — LessWrong