So, my question is about why in the cost function I needed to transpose A and not Y
{moderator edit - solution code removed}
I got the tests right, but I still don’t get it.
So, my question is about why in the cost function I needed to transpose A and not Y
{moderator edit - solution code removed}
I got the tests right, but I still don’t get it.
I don’t think you need to transpose A for this to work. What happens when you try removing the transpose for A (ie. use A instead of A.T)?
If it doesn’t work for some reason, can you do print(A.shape) and print (Y.shape) and tell us what their shapes are?
Yeah, that’s what I was thinking, it does’t really makes sense.
My sigmoid func: return 1/(1+np.exp(-z))
error when I don’t transpose A:
shapes (1,3) and (1,3) not aligned: 3 (dim 1) != 1 (dim 0)
A and Y shape are equal, that’s why the error:
(1, 3) =A.shape
(1, 3) = y.shape
mathematically it makes sense, but I don’t get my error codewise
Are you sure that error comes from the code for cost?
That specific error is more likely coming from the code for dw/db down below, where you need to call matmul and transpose accordingly.
If you’re still having issues, can you send me your code through direct message? I can help you debug.
Here’s a thread which explains the mechanics of the operations there.