Parameters to reproduce results in the paper?

[abhisheknaik96]

Thank you for releasing the codebase. The proposed approach seems quite promising! I am trying to assess the feasibility of this approach for one of my projects.

Right now I am not able to reproduce the results on n-link pendulums. The baseline with default parameters is performing slightly better than the best-performing parameters among a coarse set of parameters that I've tried for the proposed approach (over alpha, inner_epsilon, and relu smoothing).

Can you share the parameters with which you obtained the results the in the paper? Perhaps they were specified in an Appendix but I could not find any supplementary material on arXiv or the NeurIPS website.

Thanks in advance!