569ff987c643b4bedf504efda8f786c2-AuthorFeedback.pdf

Therefore, we do not13 expect that alternativelearning algorithms likeA2C orDQN would reveal interesting insights into the dynamics of14 ourenvironment. R5: You state we did a reasonably good job in describing the tasks and environment to someone unfamiliar with48 NetHack andthatwesubmitted awellwrittenmanuscript. Without specifics astowhy,wecan only say that50 we patently disagree: Throughout the paper, we make a detailed argument as to the benefits of NLE over popular51 benchmarkenvironments.