Hello, could it be a problem when grading the perceptron assignment due to random inicialization?
I just finished the assignment with apparently no errors, but the grader does not give as enterely correct two exercises because the parameter initializated randomly do not match the expected random value. I’m using the exactly same seed and also tried restarting the kernel but I never get the expected random value when initializing W
Hi @diego_iglesias!
Can you send me your Lab ID via DM so I can investigate it?
In the assignment, when you click the top right “Help” button, a panel will open and your lab ID will be shown at the bottom.
I shall take a look.
Thanks,
Lucas
Hi Lucas,
I am having the same issue as Diego. Inside the notebook, everything runs fine and code gets “All tests passed”. Also - my final costs equal the expected costs written in the markdown notes. But when I submit, grades for exercise 5 & 6 are reduced (screen shot below). Even though all grades up to there are 100%.
I will send you my notebook if that is OK?
Thank you,
Andrew.
Hi, sorry for the late response, I was OOO.
Can you send me your notebook so I can debug it? You can save and download it as .ipynb by clicking on the button in the left upper corner.
Thanks,
Lucas
Hi all,
As I can see from some notebooks that you guys have sent to me, it seems that you are using the function np.rand.rand instead of np.rand.randn in exercise 3.
It is a very small detail, but it makes a huge difference, since np.rand.randn samples from a normal distribution whereas np.rand.rand samples from a uniform one.
I hope that helps.
Thanks,
Lucas
Hi Lucas,
Thank you so much!!! 
Yep - you’re spot it - I was using rand and not randn!! Perfect score now - sorry to have wasted your time.
On the positive side - it’s proof that we are manually typing our work out which is always best 
Cheers,
Andrew.
