From State Compression to State Emphasis: Continuous Aggregation for Policy Gradient Optimization

This repository implements Hierarchical Proximal Policy Optimization (HPPO) and standard Proximal Policy Optimization (PPO) algorithms using Ray RLlib framework for reinforcement learning experiments. Requirements The project requires the following dependencies: gymnasium==1.1.1 ray==2.10.0 torch==2.0.1 Installation Install the required packages using pip: pip install gymnasium==1.1.1 ray==2.10.0 torch==2.0.1 For ALE/Atari environments, you may also need: pip install ale-py Project Structure HPPO-master/ ├── main_HPPO.py # HPPO algorithm implementation ├── main_PPO.py # PPO algorithm implementation ├── ray_result_HPPO_all.zip # HPPO training results ├── ray_result_PPO_all.zip # PPO training results └── README.md # This file