Changing the value of epsilon

#1

if self.epsilon>self.epsilon_min:

        self.epsilon*=self.epsilon_decay

shouldn’t the above code inside the act method? Why is it placed inside the train method?

#2

Hey @raunaqsingh10, no it should be in training loop only. Suppose i don’t write training code and call act function multiple times, this doesn’t mean i should decrease value of epsilon. It must only be reduced when model has learnt something and now it wants to decrease the random exploration rate.

Hope this resolved yourd doubt.
Plz mark the doubt as resolved in my doubts section. :blush: