Implementation Code for Neural Networks Paper " Representation-Driven Sampling and Adaptive Policy Resetting for Improving Multi-Agent Reinforcement Learning"
Description
This is the Implementation Code for Journal Neural Networks 's Paper " Representation-Driven Sampling and Adaptive Policy Resetting for Improving Multi-Agent Reinforcement Learning"
Files
Steps to reproduce
Installation Download the code repository using this Mendeley Data link or using: git clone https://github.com/albert-jin/eXJTU-MARL Install dependencies: pip install -r requirements.txt Usage To run the eXJTU-MARL framework, you can simply execute the following command: python train.py --config=config_file.yaml This script will initiate training based on the configurations provided. Results eXJTU-MARL has shown superior performance across various environments in SMAC, achieving higher win rates and faster convergence compared to mainstream algorithms like QMIX and VDN.
Institutions
- Xi'an Jiaotong University