Do You Need the Entropy Reward (in Practice)?