Minigrid

E438354

Minigrid is a lightweight, gridworld-based reinforcement learning environment suite commonly used for research on sample-efficient learning and generalization.

All labels observed (1)

Label Occurrences
Minigrid canonical 1

How this entity was disambiguated

Statements (50)

Predicate Object
instanceOf Python software library
gridworld environment
reinforcement learning environment suite
actionSpaceType discrete
compatibleWith OpenAI Gym interface NERFINISHED
designedFor researchers
students
focusesOn generalization in reinforcement learning
sample-efficient learning
hasAuthor Lucas Willems NERFINISHED
Maxime Chevalier-Boisvert NERFINISHED
other open-source contributors
hasDocumentation online documentation
hasEnvironment DoorKey environment
Dynamic-Obstacles environment NERFINISHED
Empty environment
Fetch environment
FourRooms environment
GoToObject environment
LavaGap environment
Memory environment
MultiRoom environment
hasFeature fast simulation
lightweight implementation
minimal dependencies
simple configurable environments
support for Gym-like API
hasLicense MIT License NERFINISHED
isBasedOn gridworld
isOpenSource true
observationType partially observable RGB image
symbolic grid encoding
programmingLanguage Python
provides benchmark environments for RL research
multiple gridworld tasks
repositoryPlatform GitHub NERFINISHED
stateRepresentation grid-based
supports curriculum learning setups
discrete action spaces
partially observable environments
procedurally generated environments
sparse reward tasks
supportsAgent single agent
usedFor algorithm benchmarking
curriculum learning experiments
exploration research
generalization studies
reinforcement learning research
representation learning research
sample efficiency evaluation

How these facts were elicited

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Gymnasium relatedTo Minigrid