Hi!
One of the questions in the Week 3 quiz from the second course in the Deep Learning Specialisation asked me to choose the recommended way to sample a value for hyperparameter alpha. I chose the option with r = -4 * np.random.rand(), but I got it wrong. I checked the lecture notes, but that’s exactly what appears there, and looking at the other possible answers, they don’t make sense. So I want to know if this is a mistake with the Coursera platform or if the lecture notes were wrong (which I doubt, since the explanation is crystal clear and makes complete sense to me).
Note: I didn’t want to write all the details of the other possible choices in the quiz for this question in order to avoid breaking forum rules.
Thanks for being careful about the forum rules! You’re right that we aren’t supposed to share quiz answers in public. But there are private ways to do that: please check your DMs for a message from me about how to do that.
Actually I just looked at my version of that quiz and I think I found the same question. The answer is not just what you showed, right? There’s another step involving exponentiation. At least if I’m really looking at the same question.
The goal of the answer is that you want a logarithmic scale for your sampled values, not a linear scale. The problem with your answer is that 10^{-4} doesn’t get you where you want to go, right?
Please take another look with the above in mind. Also let me know if I’m just looking at a different version of the question. 
Oh!!! Got it! I just counted the number of decimals! You are right, thank you so much!
1 Like
That’s great that you were able to get to the answer with that hint! Thanks for confirming.