e4d78a6b4d93e1d79241f7b282fa3413-AuthorFeedback.pdf

A strong evaluation section would compare10 training speed against other distributed / GPU-based RL systems". We believe that porting this (and similar) observation to the attention38 of the researchers can only be beneficial for the design of future simulation libraries and to develop RL algorithms39 that leverage different learning paradigms, depending on the specific data generation pattern.