Week 3 Assignment 1 Test Bug (?)

For the Two-layer-Initialization part of the exercise, I get the following output:

W1 = [[4.17022005e-03 7.20324493e-03 1.14374817e-06]
 [3.02332573e-03 1.46755891e-03 9.23385948e-04]]
b1 = [[0.]
 [0.]]
W2 = [[0.0018626  0.00345561]]
b2 = [[0.]]

This is off from the expected output (but definitely not by a constant). I have the same order of initialization as the comment of the exercise. My first guess would be a different seed, but I haven’t changed the seed that is set by the function. Is there something wrong with the test or can someone give a hint.

Kind regards
Adrian

Hi, @ahayler. Note the difference between the shapes of your matrices versus those given in the Expected output. In the latter the shapes are (4, 2) for W1, (4, 1) for b2, etc. Also, your numbers for W1 would be an awfully unlikely random draw from a standard normal distribution (although, not impossible of course!). My advice is to carefully read the instructions to Exercise 3 along with the “docstring” (function description between triple quotes) and have at it again. Also, it is helpful to post the “traceback” (i.e. error log) that your function is producing.

Yes, I agree. This platform is a medium not only to get resolve all your queries but also a medium to learn from every passing solutions provided by the mentors, super mentors, staff or anyone who contributes while working out on the solution. So, it’s a request to each and everyone out there to kindly provide the details regarding the course, the week related to the course and the exercises along with the error log (the traceback) to where you are confronting the issues. Things get easier for everyone then. Thanks!