README.md

Twin Delayed Deep Deterministic Policy Gradient (Twin Delayed DDPG or TD3)

Overview

This project focuses on implementing Twin Delay DDPG. The algorithm is implemented in Python, and is
benchmarked in various simple environments against existing implementations

Resources

Papers

• Continuous control with deep reinforcement learning - this is the original DDPG paper
• Addressing Function Approximation Error in Actor-Critic Methods - this is the TD3 Paper that addresses issues found in DDPG

Libraries

• Abseil
  • Argument parsing
• PyTorch
  • DL Framework
• Gymnasium
  • RL Environments

The Algorithm

Takeaways