The following is a DQN and Duelling DQN model trained to play the Atlantis Atari video game.
These short demonstrations take approximately 4-5 hours (Generally take 30 hours) to train each model.
To prove that learning had taken place with the DQN/ Duelling DQN a control is created with no model showing that without the models no learning takes place.
Note: This was run and tested in colab pro+.
