Deep reinforcement learning in PyTorch

This repo provides straightforward implementations of common DRL algorithms.

Algorithm list:

• DQN: Deep-Q-network
• PG: Policy gradient algorithm
• A2C: Advantage actor critic
• PPO: Proximal policy optimisation
• DDPG: Deep deterministic policy gradient
• TD3: Twin-delayed-DDPG
• SAC: Soft actor critic

The document IntroToDRL.pdf provides an introduction to deep reinforcement learning and the important formulas behind the algorithms.

Training Rewards

CartPole	CartPole

CartPole	CartPole

Pendulum	Pendulum	Pendulum

Structure

The single_file/ folder contains files with working examples of each algorithm. The modular/ folder contains the same algorithms, but split into their modular.