What Matters In On-Policy Reinforcement Learning? A Large-Scale Empirical Study

[AntoineTheb]

Do they tell how they performed that analysis. It seems an interesting aspect. @jhlegarreta

Do you mean the choices themselves or how they were grouped ? The groups were chosen purely arbitrarily based on the suspicition of the authors so there's not much to expand upon. The choices themselves were selected by looking at popular RL codebases and prior work. Unfortunately there's not much more to it, but I'll clarify.

[jhlegarreta]

Do you mean the choices themselves or how they were grouped ? The groups were chosen purely arbitrarily based on the suspicition of the authors so there's not much to expand upon. The choices themselves were selected by looking at popular RL codebases and prior work. Unfortunately there's not much more to it, but I'll clarify.

Yes, I was asking about how they chose or knew which parameters were interacting the most with each other. Your answer clarifies it. Thanks.