Implementing Dropout in a Keras DQN (reinforcement learning)

Question

First I have to say that i know, that Dropout is not common in reinforcement learning.

You can read here about the topic:

https://towardsdatascience.com/genera