RL Algorithms from Scratch

This project follows from advice on Spinning Up as Deep RL Researcher. I started by building my fundamentals in mathematics, deep learning and reinforcement learning. I studied the Deep Learning book and then followed it with David Silver’s and Sergey Levine’s course on Deep RL.

As I learned about different algorithms, I read their original papers and started reproducing them in Tensorflow and PyTorch code. Here are some of the algorithms I implemented:

• Deep Q-Network: [ code ]

• Policy Gradient (REINFORCE): [ code ]

• Asynchronous Advantage Actor-Critic: [ code ] [ notes ]

• Proximal Policy Gradient: [ code ] [ notes ]

• Soft Actor-Critic: [ code ] [ notes ]