We consider the problem of teaching via demonstrations in sequential\ndecision-making settings. In particular, we study how to design a personalized\ncurriculum over demonstrations to speed up the learner's convergence. We\nprovide a unified curriculum strategy for two popular learner models: Maximum\nCausal Entropy Inverse Reinforcement Learning (MaxEnt-IRL) and Cross-Entropy\nBehavioral Cloning (CrossEnt-BC). Our unified strategy induces a ranking over\ndemonstrations based on a notion of difficulty scores computed w.r.t. the\nteacher's optimal policy and the learner's current policy. Compared to the\nstate of the art, our strategy doesn't require access to the learner's internal\ndynamics and still enjoys similar convergence guarantees under mild technical\nconditions. Furthermore, we adapt our curriculum strategy to the setting where\nno teacher agent is present using task-specific difficulty scores. Experiments\non a synthetic car driving environment and navigation-based environments\ndemonstrate the effectiveness of our curriculum strategy.\n