Week 3 Assignment: Why there are 2 Dense layers instead of one

Course Q&A Deep Learning Specialization Sequence Models
1

Hey community!
Could somebody please elaborate why do we need 2 Dense layers instead of just one?
And why the tanh and relu were chosen as activation functions?
I tried to find the answer in the papers mentioned throughout the course but failed.

densor1 = Dense(10, activation = "tanh")
densor2 = Dense(1, activation = "relu")
2

There are two assignments in Week 3. Which one are you asking about?

3

I believe it is the first assignment, about machine translation