IRL 4/8: Maximum Entropy IRL and Bayesian IRL — LessWrong