Forager: a lightweight testbed for continual learning with partial observability in reinforcement learning
Steven Tang, Xinze Xiong, Anna Hakhverdyan, Andrew Patterson, Jacob Adkins, Jiamin He, Esraa Elelimy, Parham Mohammad Panahi, Martha White, Adam White · 2026
Read Forager: a lightweight testbed for continual learning with partial observability in reinforcement learning by Steven Tang, Xinze Xiong, Anna Hakhverdyan, Andrew Patterson, Jacob Adkins, Jiamin He, Esraa Elelimy, Parham Mohammad Panahi, Martha White, Adam White (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-01131.txt.
Description
Steven Tang and nine co-authors introduce Forager, a lightweight testbed for continual reinforcement learning under partial observability. Continual agents must keep learning, acting, and exploring in a large world that is never fully seen. Most experiments instead add non-stationarity to classic fully observable tasks and focus on loss of plasticity. Partly that is because partially seen continual environments are often prohibitively expensive to run.
Forager has a constant memory footprint and supports in-depth study of agents. Experiments show agents do exhibit loss of plasticity. Proposed mitigations can help, but leveraging state construction helps most. A variant that generates an unending stream of new tasks highlights current agents' limits clearly. That gives researchers a practical environment for memory and recurrence questions.
Catalog record
| CREATOR | Steven Tang, Xinze Xiong, Anna Hakhverdyan, Andrew Patterson, Jacob Adkins, Jiamin He, Esraa Elelimy, Parham Mohammad Panahi, Martha White, Adam White |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.01131