Giters

uvipen / Super-mario-bros-PPO-pytorch

Proximal Policy Optimization (PPO) algorithm for Super Mario Bros

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Env custom reward explain

CVHvn opened this issue · comments

commented

I see that you created a custom reward env. Can you explain why you use it and the impact of it compared with the standard reward from library?
Additionally, I see that you modified the custom reward (maybe to solve 7-4 and 4-4 levels) compared with your custom reward use in A3C.
I think you should explain the reason and importance of your custom reward. The more successful level may be because you use a better reward system not only by better algorithms?