OpenAI Gym

Cartpole-v0 using policy gradient

code

Cartpole balancing

Remain upright for as long as it can. Collect over 200 average reward per 100 episodes.

MountainCar-v0 using policy gradient

code

Mountain car

Reach to the top of the hill by limited fuel and momentum

Breakout-v0 with DQN

code

Breakout

Classical atari game of breakout.