DDPG for CarRacing Using Deep Deterministic Policy Gradient to solve CarRacing from OpenAI gym. Unfortunately the CarRacing environment / Box2D has a memory leak bug (as of 1/19/21) which prevents the agent from training effectively.