Possible Error in Building_a_Recurrent_Neural_Network_Step_by_Step

Course Q&A Deep Learning Specialization Sequence Models
1

imagem
imagem1038×423 13.4 KB

Should x_tmp.shape[-1] == da_tmp.shape[-1] ?

2

The shapes are correct. x_tmp.shape[-1] != da_tmp.shape[-1]

3

I thought that last dimension was Tx, which should be the same between input and activation?

imagem
imagem962×628 59.6 KB

4

You’re right about the general observation. In the test, we’re performing the calculations for the first 4 steps. If you look at the comment
#Retrieve the dimensions from da's and x1's shape, take T_x from da.shape[-1]