An implementation of DeepMind's Deep Q-Network (DQN) for playing Atari games. The first DQN is modeled for the simpler CartPole-v1 environment, which was used for testing and evaluation. The second DQN is an altered version of the first DQN modeled after DeepMind's DQN for playing Atari games, and was trained for 5000 episodes over ~6 hours on TPU to play Pong.
- Emir Esenov
- Ola Karrar