This repository provides distributed implementations of deep neuroevolution algorithms, including Evolution Strategies (ES), Novelty-Seeking ES (NS-ES), and a Deep Genetic Algorithm (DeepGA). It targets researchers and practitioners in reinforcement learning and evolutionary computation, offering a competitive alternative to gradient-based methods for training deep neural networks.

How It Works

The project implements distributed training of neuroevolutionary algorithms, enabling parallel exploration of policy spaces. It leverages a population-based approach where multiple agents evolve concurrently, facilitating efficient search in complex reinforcement learning environments. The implementation is designed for scalability, supporting local execution and cloud deployment on AWS.

Quick Start & Requirements

• Install: Clone the repository, create a Python 3 virtual environment, activate it, and install requirements with pip install -r requirements.txt.
• Prerequisites: Python 3, Redis, Mujoco (for Humanoid experiments, requires a separate license and binary).
• Running Experiments: Use scripts/local_run_exp.sh with specific configuration files (e.g., es configurations/frostbite_es.json).
• Visualization: Run policies using python -m scripts.viz <ENV_NAME> <YOUR_H5_FILE>.
• Docker: Instructions are provided for running within Docker containers.
• Docs: Visual Inspector for NeuroEvolution (VINE) and GPU Implementation have separate READMEs.

Highlighted Details

• Implements ES, NS-ES, NSR-ES, and DeepGA.
• Supports Atari (Frostbite) and MuJoCo (Humanoid) environments.
• Includes VINE for interactive visualization of neuroevolution results.
• Offers a GPU-optimized implementation for enhanced performance.

Maintenance & Community

The project originates from Uber AI Labs. Links to related papers and OpenAI's original code are provided. Community interaction channels are not explicitly mentioned in the README.

Licensing & Compatibility

The repository's license is not explicitly stated in the README. Compatibility for commercial use or closed-source linking would require clarification of the licensing terms.

Limitations & Caveats

The Humanoid experiment requires a separate Mujoco license and binary installation. The project's current maintenance status and community activity are not detailed.