Hi, here my results
Ex1
J = 8
All tests passed.
Ex2
dtheta = 3
All tests passed.
Ex3
3 2.9999999995311555
Your backward propagation works perfectly fine! difference = 1.221195460051932e-20
Ex4
Your backward propagation works perfectly fine! difference = 2.827876250996049e-14
But I got this error:
AssertionError Traceback (most recent call last)
in
6 expected_values = [0.2850931567761623, 1.1890913024229996e-07]
7 assert not(type(difference) == np.ndarray), “You are not using np.linalg.norm for numerator or denominator”
----> 8 assert np.any(np.isclose(difference, expected_values)), “Wrong value. It is not one of the expected values”
AssertionError: Wrong value. It is not one of the expected values
1 Like
It looks like you are making a mistake in computing the difference in both cases. Here’s what I get as the answer for Ex3:
Your backward propagation works perfectly fine! difference = 2.919335883291695e-10
Notice that in both cases, your answer is about the square of the correct answer at least in terms of order of magnitude. BTW in Ex4, I assume you had already fixed the intentional bugs that they put into the back prop function.
I suggest you carefully compare your implementation of the difference calculation to the math formulas given in the assignment and the lecture. Are you sure you didn’t use the square of the norm in the numerator?
1 Like
Note that in the formulas when they say ||v||_2, that’s a “subscript 2” there, meaning that is the 2-norm. It’s not saying to square the norm. That would look like this: ||v||_2^2.
1 Like
Thanks for the answer, I got it.
1 Like