eba237eccc24353ccaa4d62013556ac6-AuthorFeedback.pdf

However, properly evaluating theγ-dependent behavior for the38 non-linear case is non-trivial. The main reason for this is that DQN contains a lot of hidden hyper-parameters that39 work well forγ = 0.99,butit'sunclear ifthese are also agood choice for differentγ-values.