Unclear instructions when disentangling the noise vectors

When disentangling the noise vectors shown here:

class MappingLayers(nn.Module):
    Mapping Layers Class
        z_dim: the dimension of the noise vector, a scalar
        hidden_dim: the inner dimension, a scalar
        w_dim: the dimension of the intermediate noise vector, a scalar
    def __init__(self, z_dim, hidden_dim, w_dim):
        self.mapping = nn.Sequential(
            # Please write a neural network which takes in tensors of 
            # shape (n_samples, z_dim) and outputs (n_samples, w_dim)
            # with a hidden layer with hidden_dim neurons

The last comment before the code section is misleading on the vector dimensions to include as part of the appropriate layers. I’d like to clarify what the appropriate dimensions should be and then update the notebook accordingly.

@Bradley_M_Messer, definitely if you have a suggestion for a version of the comment that is more clear, just post your suggestion here, and I can pass it along to the assignment developers, who can change the master copy of the assignment so all students can take advantage of the improved comment.
Keep in mind, though, that we don’t want to give TOO much away in the comment. We want to let the student start to figure out how to apply what they’ve learned so far, so they’ll be able to do this on their own when they finish the course. For those students who want or need a little more direction, we also have the optional hints they can look at.

@Wendy Thank you for your taking time to respond. I’ll look it over and modify it to something that allows maximal flow for the least investment without giving away the answer. The structure of the comment iself threw off an otherwise easy graded section

@Wendy Luckily context clues presented in the test cell towards the bottom were sufficient to help work through the confusion. I’d just also add a reminder that lots of us work as professionals, lots of mentoring, greenfield projects, house owning etc, with 2.5 books read separately while still taking this course with few days off.

Having the thorough education like what I’ve presented here:

to help people get up to speed is always appreciated. Separately, I want to mention how much I appreciate deeplearning.ai’s help in getting this far though is appreciated though. My situation wasn’t ideal and this team helped pull me out of that. I’m hoping this can be considered payment for all the good you’ve done for me :slight_smile:

1 Like

My apologies by the way, I felt very overwhelmed at this moment with everything going on. This will just help us get up to speed faster. Thanks!

No problem, @Bradley_M_Messer - completely understand how hectic things can get sometimes. Thank you also for your kind words about the help you received from the deeplearning.ai team!

Thank you Wendy as you could have reacted much differently than that. My 7 years of industry experience has really evolved my understanding of the role that education should play. That’s mainly that if I solve my own problems, I should open source the solution as much as is reasonable because that enables other people to solve their problems much faster and in today’s digital age, that’s a massive competitive differentiator.

I’ve also realized since I study so much 630 hrs in just this past year, skills are fleeting at best. Some of these skills won’t matter in 2 years-4years etc. I hope that’s not the case, but things just evolve so fast to be too optimistic that my skillset will be any more stable. That’s not to say I won’t try, but the digital age and competitive differentiator demand a different perspective than the traditional these skills will last me my natural lifetime (because they just won’t and then you’re be scrambling to find a new job when something happens.).