PettingZoo VS open_spiel

Release notes: https://github.com/Farama-Foundation/PettingZoo/releases/tag/1.24.0

To address this issue, we are excited to announce the release of Shimmy as a mature Farama Foundation project. Shimmy is an API compatibility tool for converting external RL environments to the Gymnasium and PettingZoo APIs. This allows users to access a wide range of single and multi-agent environments, all under a single standard API.

I am currently trying to implement my own version of a Connect Four Environment based on the version available on the PettingZoo Library github (https://github.com/Farama-Foundation/PettingZoo/blob/master/pettingzoo/classic/connect_four/connect_four.py).

The Farama Foundation is always looking for contributors to PettingZoo, the largest open source MARL library out there (https://github.com/Farama-Foundation/PettingZoo, farama.org). If you might be interested in that you can message them in their discord server

I am trying to implement a custom PettingZoo environment, and a shared policy with Stable Baselines 3. I am running into trouble with the action spaces not being compatible, since PettingZoo has started using gymnasium instead of gym. Does anyone know if these libraries no longer work together, and perhaps if there is a work-around?

I'd say this is a great path but I'd also look at the basic on-policy gradient actor critic methods like A2C and eventually PPO. Someone recommended SAC which also really good. There are tons of environments in the https://github.com/Farama-Foundation/PettingZoo as well if you want to mess with those. You can also check out stable baselines https://github.com/DLR-RM/stable-baselines3 which is pretty popular. If you want to get into the theory more I recommend reading the Sutton and Barto book on reinforcement learning.

Might want to check out petting zoo: https://github.com/Farama-Foundation/PettingZoo

Not that I know of, most I've seen are based on gym. I did see PettingZoo which is for multi-agent RL. https://github.com/Farama-Foundation/PettingZoo

Thank you for your comment. This is certainly helpful. I have since read more about this domain and turns out mathematically it is modeled using Markov/Stochastic game representations. I also found PettingZoo which can be used with RayLib for MARL problems.

Deep mind actually has a repository where they applied this algorithm for incomplete-knowledge games. You could use it for reference: https://github.com/deepmind/open_spiel/tree/master/open_spiel/python/algorithms

+1. You can also find an implementation of Clobber and AlphaZero (and many other basic RL algorithms) in OpenSpiel: https://github.com/deepmind/open_spiel

You can find a lot of code online if you look, and many of these applied to Poker. There's a general implementation of both in Python and C++ in OpenSpiel, with some examples applied to small poker games. It's nice code to learn from because the algorithms operate over generic game descriptions, so there aren't game-specific design choices mixed up with the implementation of the algorithms, and you can create your own poker game and just run them on it.

And many other additions and improvements. See all the details here: https://github.com/deepmind/open_spiel/releases/tag/v1.3

I would checkout the openspiel package. It's main focus is RL in games (multi-agent environments). You'll find RL examples there and games that are small enough to solve without deep RL. There's also a wide range of environments from fully cooperative to adversarial zero-sum.

Hi, you can check out OpenSpiel: https://github.com/deepmind/open_spiel/

as for algorithms , openspiel repository has few implementations some of these are not related to imperfect information games , and others are not for multiagent environment and others are tabular algorithms .

This includes single-agent Gymnasium wrappers for DM Control, DM Lab, Behavior Suite, Arcade Learning Environment, OpenAI Gym V21 & V26. Multi-agent PettingZoo wrappers support DM Control Soccer, OpenSpiel and Melting Pot. For more information, read the release notes here:

I've had some success using Action Masking - you can refer to here https://github.com/deepmind/open_spiel/blob/120420a74a69354d64c10b51cd129d4587f9f325/open_spiel/python/algorithms/dqn.py but for DQN you need to mask out q values for invalid actions (as well as masking them during prediction). In my case I'm able to place my mask in the observation so can fetch it quite easily during prediction but if that's not possible you could query it from the environment and store it in the replay buffer (like they do in the link I shared)

Take a look at this simple implementation: https://github.com/deepmind/open_spiel/blob/master/open_spiel/algorithms/minimax.cc