Diversity Evolutionary Policy Deep Reinforcement Learning

Table 2

Values of hyperparameter.

Hyperparameter	Values
Critic/actor learning rate	0.0003
Critic/actor hidden layer	2
Number of neurons	400/300
Critic activation	Relu
Actor activation	Tanh
Discount factor	0.99
Optimizer	Adam
Soft update coefficient	0.005
Experience pool capacity	10^6
Experience pool sample size	100
Gauss noise	Clip ((0, 0.2), −0.5, 0.5)