Abstract: Most multiagent reinforcement learning (MARL) methods are limited in the scale of problems they can handle. With increasing numbers of agents, the number of training iterations required to ...
🎉 [2025-12] MARIE now supports SMAX (JAX-based StarCraft Multi-Agent Challenge) with 14 available maps! We provide CPU-only JAX integration to ensure compatibility with Ray distributed workers. See ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback