Gymnasium
E95192
Gymnasium is a modern open-source toolkit for developing and comparing reinforcement learning algorithms, serving as a successor and refinement to OpenAI Gym.
Statements (49)
| Predicate | Object |
|---|---|
| instanceOf |
open-source project
→
reinforcement learning toolkit → software library → |
| aim |
improve clarity of environment termination semantics
→
provide a maintained successor to OpenAI Gym → standardize benchmarking of RL algorithms → |
| basedOn |
OpenAI Gym
NERFINISHED
→
|
| compatibleWith |
JAX
NERFINISHED
→
NumPy → PyTorch NERFINISHED → TensorFlow NERFINISHED → |
| developer |
Farama Foundation
NERFINISHED
→
|
| documentation |
https://gymnasium.farama.org
→
|
| ecosystem |
Farama Foundation reinforcement learning ecosystem
→
|
| feature |
compatibility with existing Gym environments
→
registration system for custom environments → seeding utilities for reproducibility → standardized API for RL environments → step API with termination and truncation signals → type-annotated API → vectorized environment interfaces → wrappers for environment preprocessing → |
| genre |
AI research
→
machine learning → reinforcement learning → |
| hasAPI |
Env interface
→
VectorEnv interface → Wrapper interface → |
| license |
MIT License
NERFINISHED
→
|
| maintainer |
Farama Foundation
NERFINISHED
→
|
| name |
Gymnasium
→
|
| openSource |
true
→
|
| predecessor |
OpenAI Gym
NERFINISHED
→
|
| programmingLanguage |
Python
→
|
| relatedTo |
Gymnasium-Robotics
→
Minigrid → PettingZoo NERFINISHED → |
| repository |
https://github.com/Farama-Foundation/Gymnasium
→
|
| supports |
Atari environments via wrappers
→
Box2D environments → classic control environments → continuous action spaces → discrete action spaces → toy text environments → vectorized environments → |
| supportsVersion |
Python 3.8+
→
|
| usedFor |
benchmarking reinforcement learning algorithms
→
developing reinforcement learning algorithms → teaching reinforcement learning → |
Referenced by (4)
| Subject (surface form when different) | Predicate |
|---|---|
|
RLlib
→
Stable Baselines → |
supportsEnvironmentInterface |
|
OpenAI Gym
→
|
hasSuccessor |
|
OpenAI Gym
→
|
influenced |