Randomness in optimal epsilon_greedy_policy
levindabhi opened this issue · comments
Levin Dabhi commented
Why we are not decaying epsilon in epsilon_greedy_policy ?
Apoorv Agnihotri commented
@levindabhi I am not sure but isn't it a type of a hyperparameter? If you would like to decay it, then you can else you can just assume that the minimum \epsilon allowed is the \epsilon you gave to the algorithm in the first place.