This repository provides implementations of various reinforcement learning algorithms using TensorFlow 1.0, targeting researchers and students interested in deep RL. It offers educational code examples for learning and experimenting with RL models, evaluated against OpenAI Gym environments.

How It Works

The project implements classic and deep reinforcement learning algorithms, including Cross-Entropy Method (CEM), Q-Learning, Deep Q-Networks (DQN), Double DQN, REINFORCE, Actor-Critic, and Deep Deterministic Policy Gradient (DDPG). Models are built using TensorFlow 1.0, leveraging its graph computation capabilities for efficient training and inference.

Quick Start & Requirements

• Install: pip install -r requirements.txt
• Prerequisites: TensorFlow 1.0, Python 2.7 or 3.5, OpenAI Gym.
• Links: OpenAI Gym

Highlighted Details

• Implements a range of RL algorithms from tabular methods to policy gradients.
• Models are evaluated in standard OpenAI Gym environments.
• Codebase supports both Python 2.7 and 3.5.

Maintenance & Community

The project is maintained by yukezhu. Contributions and feedback are welcomed.

Licensing & Compatibility

• License: MIT
• Compatibility: Permissive MIT license allows for commercial use and integration into closed-source projects.

Limitations & Caveats

The implementations are primarily for educational purposes and may require modifications to work effectively on specific RL problems. The project uses TensorFlow 1.0, which is an older version and may not be compatible with newer TensorFlow ecosystems or hardware acceleration features.