Gradient_Checking assignment returns unknown error

The function gradient_check_n_test_case() returns a test error. I’ve checked this link and this link. I updated the notebook and wrote my code again and check all of my implementations and all of them are correct. What’s my issue :pensive:

Well, there are some other things besides the issues on those two links. Another one I can think of is “order of operations” on the computation of grad_approx because of the division by 2\epsilon. As an example of what I mean, try this experiment and watch what happens:

m = 5.
x = 1. / 2. * m
y = 1. / (2. * m)

You might expect x and y to be the same, but they’re not. :nerd_face:

Another thing to check is whether the grad check value is the same as the expected value that they show in the case that you have not yet fixed the intentional bugs that they put into back propagation. With my code, I get exactly the value they show. If you don’t, then there is something different in your code that you need to find.