All Tools
E
AgentsFreeOpen Source
EPYMARL
Extended PyMARL for fair multi-agent RL benchmarks
Apache-2.0
ABOUT
Comparing MARL algorithms is hard when each paper ships a different stack. EPyMARL keeps PyMARL's training loop and adds MAPPO, IPPO, Gymnasium envs, and shared vs unshared parameters so cooperative agent results stay comparable across LBF, RWARE, and SMAC.
INTEGRATION GUIDE
1. Benchmark MAPPO, QMIX, and IPPO on Level-Based Foraging or RWARE
2. Train cooperative agents on SMAC and SMACv2 StarCraft maps
3. Run PettingZoo or VMAS environments through a shared MARL trainer
4. Log multi-agent experiments to Weights and Biases for comparison
TAGS
pythonagentsmulti-agentreinforcement-learningmarlpytorchgymnasium