The DOI matches its DataCite record.
Gymnasium
Open accessAbstract
An API standard for single-agent reinforcement learning environments, with popular reference environments and related utilities (formerly Gym)
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Current reinforcement learning or control theory frameworks such as Gymnasium do not give end-users fine-grained control over details of the environment, instead focusing on implementing a single agent-environment interaction through a transition function.
“Current solutions from the reinforcement learning or control theory community, such as Gymnasium [23] do not give end-users these controls over details of the environment, instead focusing on implementing a single agent-environment interaction through a transition function.”
From Reactive Environments for Active Inference Agents with RxEnvironments.jl (2025) · cites this paper as [23]
In popular environment design frameworks such as Gymnasium, the transition function for an environment takes the action emitted by an agent and the previous state, and produces a new state and an observation for the agent.
“In popular environment design frameworks, such as Gymnasium [23], the transition function for an environment takes the action emitted by an agent and the previous state. It produces a new state and an observation for the agent.”
From Reactive Environments for Active Inference Agents with RxEnvironments.jl (2025) · cites this paper as [23]
Citations
Cited by
Sorted by citations
Reactive Environments for Active Inference Agents with RxEnvironments.jl
Nuijten, W. W. L., & de Vries, B. (2025) · Communications in computer and information science · doi:10.1007/978-3-031-77138-5_10
Metadata
- Authors
- Towers, M., Terry, J. K., Kwiatkowski, A., Balis, J. U., Cola, G., Deleu, T., Goulão, M., Kallinteris, A., KG, A., Krimmel, M., Perez-Vicente, R., Pierré, A., Schulhoff, S., Tai, J. J., Tan, A. J. S., & Younis, O. G.
- Affiliations
- University of Southampton (GB)
- Farama Foundation
- University of Wisconsin–Madison (US)
- Université de Montréal (CA)
- Instituto Superior Técnico (PT)
- Technical University of Crete (GR)
- connectome.design
- ETH Zurich (CH)
- Brown University (US)
- University of Maryland, College Park (US)
- Coventry University (GB)
- University of Bologna (IT)
- ORCID
- Venue
- Zenodo
- Publisher
- Zenodo
- Published
- 2024-02-13
- OpenAlex
- W6968906521
- Cited by
- 77
- References
- 0
- Citations by year
- 2026: 11 · 2025: 18 · 2024: 42
- Type
- dataset
- Language
- en
- Open access
- Licence
- other-oa
- opensource.org/licenses/MIT
- Locations
- ePrints Soton (University of Southampton) (acceptedVersion)
- Pure (Coventry University) (submittedVersion, other-oa)
- Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, other-oa)
- Zenodo (CERN European Organization for Nuclear Research)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- gymnasium, reinforcement learning environment, interoperability, benchmark environments, open-source library, software design decisions, training algorithms
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
