PacktPublishing / Deep-Reinforcement-Learning-Hands-On

Hands-on Deep Reinforcement Learning, published by Packt

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Current versions for OpenAI `gym` and other packages -- `PongNoFrameskip-v4` is not found in current `gym`, `atari-py`

00krishna opened this issue · comments

Hello Folks.

I was just wondering how to adapt the repo code to the current versions of gym and atari-py. It seems that a lot of the classic
RL control environments were migrated to the ALE packages and some other things. I tried to use the PongNoFrameskip-v4 environment, but I kept getting an error message about the environment not being registered or indexed--even though I did install the ROMs before, etc. I was testing out the new Pong-v5 or something environment to see if that will work instead, but still just trying to get it to work.

Any suggestions would be appreciated. Of course trying to keep up with all of the versions is a full time job, right. But it seems like so many of the other tutorials out there are based upon the older version of OpenAI as well, so no luck finding other implementations to emulate.

Thanks.

Hi I am facing the same issue. Did you manage to find some solution?

@dayemsiddiqui Yeah, I believe that this environment has been moved to a new package called ale-py. I have not tried it myself, as I was using the Pong-v5 or something version from gym. But someone told me that these legacy environments were moved to the new package. I know the documentation was not really updated anywhere.