-
Maximum-Entropy RL: The 1-Step Problem
Working through why an entropy bonus turns argmax into softmax and max into log-sum-exp, step by step, then following it into soft value iteration, SAC and robust RL.
Nothing here on this topic yet. See all posts.