Deep RL Agents for Atari problems Deep RL agents tested on the Pong game Methods Dueling DDQN (model_state contains a network pre trained for ~200K frames)