An API conversion tool for reinforcement learning environments.

Shimmy provides Gymnasium and PettingZoo bindings for popular external RL environments.

DM Control

DM Control: 3D physics-based robotics simulation.

DM Soccer

DM Control Soccer: Multi-agent cooperative soccer game.

DM Lab

DM Lab: 3D navigation and puzzle-solving.

Behavior Suite

Behavior Suite: Test suite for evaluating model behavior.

Atari Learning Environment

Atari Learning Environment: Set of 50+ classic Atari 2600 games.

Melting Pot

Melting Pot: Multi-agent social reasoning games.

OpenAI Gym

OpenAI Gym: Compatibility support for Gym V21-V26.

OpenSpiel

OpenSpiel: Collection of 70+ board & card game environments.

Environments can be interacted with using a simple, high-level API:

import gymnasium as gym
env = gym.make("dm_control/acrobot-swingup_sparse-v0", render_mode="human")

observation, info = env.reset(seed=42)
for _ in range(1000):
   action = env.action_space.sample()  # this is where you would insert your policy
   observation, reward, terminated, truncated, info = env.step(action)

   if terminated or truncated:
      observation, info = env.reset()
env.close()