I tried to run a Double Dueling Deep Q-learning Network (DDDQN) that is supposed to learn how to play Doom and understand that he needs to kill enemies before being able to move