Curiosity Gym is a unified Python framework designed to develop, evaluate, and compare curiosity-driven reinforcement learning algorithms. The framework provides three predefined Gymnasium environments, each of which has been designed to address a specific goal of curiosity-driven approaches. In addition to its focus on the distinctive characteristics of curiosity-driven RL, Curiosity Gym emphasizes usability and flexibility, making it an accessible tool for both researchers and practitioners in the field. The documentation website can be found here.
Curiosity Gym requires Python 3.12 or greater.
git clone https://github.com/chrisreimann/curiosity-gym
cd curiosity-gym
pip install .
To use the predefined environments from the Curiosity Gym framework, simply import SparseEnv, DistractiveEnv or MultitaskEnv from the package. For Example:
# Import predefined environment
from curiosity_gym import SparseEnv
# Initialize environment
env = SparseEnv(agentPOV="local_2", render_mode="human")
# Use Gymnasium API
observation, info = env.reset()
action = env.action_space.sample()
observation, reward, terminated, truncated, info = env.step(action)
env.close()
After an environment is initialized, all Gymnasium API methods, like step, reset or close can be used.