Hi, I’m a linguistics PhD student and I’m interested in NLP, which is not something that anyone does in my dept. I have taken a couple semesters of statistics as a grad student and I have used R (but I’m not an expert at all - I always find code and modify it for my needs). I followed the logic of the lectures just fine, and I completed and mostly understood the practice labs. I do not know any calculus, so the math was above my head, but the lecture notes said that didn’t matter.
I have been following along with the assignment lab preliminaries (I would never have guessed how to write the code for that, but as I read it, I mostly get what’s going on) and now I’m stuck on part 1.1. I don’t know what I should add or change - I’m assuming I am supposed to add and modify code, but since I can’t find any code like it in the practice labs, I have absolutely no idea what I am supposed to do. I work best with examples that I can fix or modify. I also don’t even completely understand the equation for the sigmoid function. I do not at all feel prepared by the lectures and practice labs to do this lab. The lectures say I’ve learned how to do this now that I’ve gone through the material, but clearly not. I am not able to make the leap that apparently most students here are able to make in order to complete this code. I probably need to complete some Python courses first, at the very least. I don’t think the one beginner course offered by DeepLearning.AI would be enough, since it is supposed to be nontechnical.
I’m pretty disappointed, since the topic is interesting and I would love to be able to run cool analyses on my own linguistics data someday! I would appreciate it if anyone has any suggestions for resources that would help me figure this out. If I can’t, then I’ll need to cancel my subscription to this specialization before the end of the week is up. Thanks.
I am so sorry to hear about your frustrations, but honestly, it makes me feel a bit better at the same time (if that makes sense). I have actually progressed a little bit further in the assignment (I’m partway through 1.2), but only because my partner, who is more familiar with Python, was helping me. I have no idea which functions exist that I can use and I’m not familiar with the way the syntax works - there are similarities to R, but it’s not quite the same, so I’m tripping up. From your comment it sounds like I’ll have more issues beyond just not being familiar with Python! That’s good to know. Thanks so much for your comment.
Hi rfritche and David_Simmonds,
On the deeplearning.ai website, the NLP course is presented as a course of intermediate complexity. In my opinion, it could also have been presented as advanced. The (eventual) focus on transformers and the use of trax make it quite a challenging specialization. It is also clear that a good working knowledge of python is a prerequisite.
To get a good working knowledge of python I would suggest considering taking the first three or four courses of the excellent Python 3 Programming Specialization at Coursera taught by faculty of the University of Michigan. You can find that specialization here.
To get a full understanding of the NLP specialization, it may also be useful to first take the Deep Learning Specialization at deeplearning.ai.
So yes, it’s challenging, it requires an investment, and it may not suit everyone. On the positive side, the specialization takes you straight from a starting level in NLP to the most advanced NLP models, so there is a considerable payoff at the end. Also, improvements are constantly made to the provided courses. Hopefully, your feedback will help with further improvements to the specialization and the way it is presented on the deeplearning.ai website.
So for what it is worth, thanks for your feedback and good luck to you on your learning journey.
Thank you reinoudbosch I am sure I will figure it out.
You will go through this!
While I do think the course is definitely an intermediate one … I think the synchronization in messaging between the content/videos and the labs is very poor. Things that are “optional” in the videos are “must understand” in the labs and even in the practices tests.