Due to the variety of methods and frameworks used to train agents, we here provide wrappers for most of them and showcase the flexibility provided by imp_marl.
The wrappers simply use the interface defined in imp_env.py in order to plug any IMP environment into any framework.
Examples of the use of these wrappers can be found in examples.
Each wrapper has its own requirements. Please check the import of each wrapper, where the requirements are listed.
- PymarlMAStruct: Multi-agent wrapper for the MultiAgentEnv interface required by PyMarl.
- PymarlSAStruct: Single-agent wrapper for the MultiAgentEnv interface required by PyMarl.
- EPymarlMAStruct: Multi-agent wrapper for the interface required by EPyMarl.
- GymSaStruct: Single-agent wrapper for the Gym v21 interface.
- GymnasiumSaStruct: Single-agent wrapper for the Gymnasium interface.
- PettingZooStruct: Multi-agent wrapper for the PettingZoo interface.
- MarllibImpMarl: Multi-agent wrapper for MARLlib.