CARVIEW

MOTORHOMES

Select Language

HTTP/2 301 server: GitHub.com content-type: text/html location: https://marcometer.github.io/jmlr_2024.github.io/ x-github-request-id: 8A39:2680BD:963954:A8BB5F:69531AED accept-ranges: bytes age: 0 date: Tue, 30 Dec 2025 00:21:02 GMT via: 1.1 varnish x-served-by: cache-bom-vanm7210024-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1767054062.117771,VS0,VE200 vary: Accept-Encoding x-fastly-request-id: 2016c551aac712fc8fcb8267b61276a3030b4ba7 content-length: 162 HTTP/2 200 server: GitHub.com content-type: text/html; charset=utf-8 last-modified: Fri, 28 Feb 2025 06:21:33 GMT access-control-allow-origin: * etag: W/"67c155ed-803d" expires: Tue, 30 Dec 2025 00:31:02 GMT cache-control: max-age=600 content-encoding: gzip x-proxy-cache: MISS x-github-request-id: D552:3A7A40:964085:A8C47C:69531AEC accept-ranges: bytes age: 0 date: Tue, 30 Dec 2025 00:21:02 GMT via: 1.1 varnish x-served-by: cache-bom-vanm7210024-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1767054062.331075,VS0,VE227 vary: Accept-Encoding x-fastly-request-id: b4a79b2aae112e62658b2514af55af0758f48d7c content-length: 3862 Memory Gym Experiments

Memory Gym: Towards Endless Tasks to
Benchmark Memory Capabilities of Agents

Memory Gym features the environments Mortar Mayhem, Mystery Path, and Searing Spotlights. These environments benchmark an agent's memory to

memorize events across long sequences,
generalize,
and be robust to noise.

Especially, these environments feature endless task variants. As the agent's policy improves, the task goes on. The traveling game "I packed my bag ..." inspired this dynamic concept, which allows for examining levels of effectiveness instead of just sample efficiency.

Github

Memory Gym neroRL (Deep Reinforcement Learning Framework) Recurrent Proximal Policy Optimization using Truncated BPTT Transformer-XL as Episodic Memory in Proximal Policy Optimization CleanRL Implementation of PPO-TrXL Evaluation Results and Jupyter Notebooks for Plotting the Paper's Figures

Citation

@article{JMLR:v26:24-0043,
     title={Memory Gym: Towards Endless Tasks to Benchmark Memory Capabilities of Agents},
     author={Marco Pleines and Matthias Pallasch and Frank Zimmer and Mike Preuss},
     journal={Journal of Machine Learning Research},
     year={2025},
     volume={26},
     number={6},
     pages={1--40},
     url={https://www.jmlr.org/papers/v26/24-0043.html}
}

Interactive Result Videos of Trained Agent Behaviors

For each environment, we cherry picked GRU and TrXL agent behaviors for visualization.
Three episodes, related to level seeds not seen during training, were rendered for every chosen agent.
Clicking on distinct data points in line plots or attention scores will take you to the corresponding episode step.
Each of the upcoming result pages features various visualization tools.