GenEnv-MARL Integration Project

# GenEnv-MARL Integration Project ## Description This project explores the integration of GenEnv with multi-agent reinforcement learning (MARL) frameworks to enhance coordination and cooperation among large language model (LLM) agents in complex, dynamic environments. ## Research Hypothesis Integrating GenEnv with MARL frameworks will enhance coordination and cooperation among LLM agents in complex, dynamic environments. ## Implementation Approach - Develop a MARL framework integrated with dynamic GenEnv environments. - Design tasks requiring cooperation among agents to achieve shared goals. - Use performance metrics to evaluate improvements in coordination. - Compare results with baseline MARL setups without co-evolutionary environments. ## Setup Instructions 1. Clone the repository: ```bash git clone https://github.com/yourusername/GenEnv_MARL_Project.git cd GenEnv_MARL_Project ``` 2. Install the required packages: ```bash pip install -r requirements.txt ``` ## Usage Examples - To train the agents, run: ```bash python src/train.py ``` - To evaluate the trained models, run: ```bash python src/evaluate.py ``` ## Expected Results We expect to observe improved cooperation among LLM agents in GenEnv-integrated MARL setups compared to traditional MARL environments. ## References - [GenEnv: Difficulty-Aligned Co-Evolution Between LLM Agents and Environment Simulators](http://arxiv.org/abs/2512.19682v1)