Assignments were extremely didactic; there was no room for creativity. They were not transparent and gave a minimal idea of how to implement these things properly. Course moderators did not bother to answer any of my queries, making the course even less intellectually stimulating. The lectures were monotonous, and hence, I was having trouble finding them to be very engaging. Although, the professor did give some insightful points.

In conclusion, I wouldn't recommend this course to someone unless they are extremely novice programmers. Yet, one may refer to the videos to gain some conceptual clarity on specific topics.

Unfortunately considerably a subpar course compared to the other four in the specialization. Programming assignment is a mess - wrong formulas presented, nowhere near enough Keras's tutorials, etc. Every assignment is passed by browsing the forum looking for help from other people. It is unclear to the point of being annoyed (got someone in the forum cancel his subscription). However, lectures are fine and sequence models cover a wide range of areas/applications, so you can't miss it anyway.

This course felt rushed. Especially, the programming assignments, which had many errors and were frustrating at time. It is still worth it since the content is really good -- only if you are willing to go through the frustration during the programming exercises.

Assignments don't match the quality of the other four courses of this specialization. Automatic grading accepts solutions despite results not matching expected results. This should be fixed.

This one went a little fast for me, can't say that I'm confident on the shapes of tensors going through RNNs and why

Some coding assignments were too hard to follow what was required.

This course has many inconsistencies and errors in the homework. Seems like a rushed job.

some optional exercises are wrong, wasted a lot of time on LSTM backward propagation

It was difficult to follow the programming exercises because many of it had already been written. I think it would be more useful to learn one framework instead of using both keras and tensorflow in one course. I still don't know how to debug any of these frameworks. Without the forums, it would be very difficult to pass the assignments. Sometimes there were bugs in the jupyter notebook, sometimes typos that were misleading. As a result, it took me many hours stuck on one assignment. It would be good if these comments are taken into account for the future classes of this course. I really enjoy Andrew Ng.'s courses but I was disappointed at this last course's assignments.

The unit tests in the programming assignments are poorly implemented. They will fail you if your code is not exactly as expected, even when it runs and returns the correct output.

TensorFlow and Keras basics are not covered. The course states no pre-requisites as well. This was really disappointing.

There are a lot of mistakes in programming assignment. Please update and fix it

Wonderful end to this Deep Learning Specialization. The programming assignments cover up a variety of hot topics in the Deep Learning market. The videos are very well made and teach the content in depth. A special thanks to Prof. Andrew for yet another amazing course in this wonderful specialization!

Great module, I am lucky to have used this resources in learning sequence models, I can imagine running LSTM using one of the frameworks without ever implementing it myself, Andrew Ng/Deeplearning.ai is the best!

Amazing course. Andrew Ng has exceptional talent to explain complicated concepts. I have heard about RNNs in other courses but this is the first course, that actually made me understand them. Highly recommended.

Best simple course for Deep Learning. I think this specialization is the best as a MOOC but it can be better as an academic course.

Professor Andrew is really knowledgeable. I learn a lot from his lecture videos.

Very good course which gives a nice intuition to sequence deep learning modelling. Unfortunately, this is the weakest one among the whole specialization. There are no deep explanation of LSTM as well as GRU and back-propagation algorithm. Seq2seq models explanation is not clear and looks too inconsistent. I had to read a lot of the additional materials and blogs in order to understood a theory behind lectures. Hence, the first week assignments were disagreeably difficult to complete, whereas second and third week assignments were comparatively easy. I think this course should be revised or prolonged for 4 weeks to cover LSTM models more profoundly. Nevertheless, I would like to thank Prof. Andrew Ng for really great job and initiatives in such an important area of study!

Little bit math heavy. It was sometimes hard to understand the intuition, e.g. RNN, LSTM, GRU

Could have been more thorough like previous courses

I would really like to give this course 5 * but the finally programming assignment was a disappointment. It seems many other folks feel the same way. I found myself spending many hours trawling the the web for additional background.

The first week is amazing. The last week is the worst! Andrew starts nicely; but as he goes to the second and third weeks, he hardly explains why he does what he does.

Week 4 coding exercise is incomparably harder than any other in this entire specialization.

The transformer subclass programming exercise is super useless task. Spent hours on this task and learnt nothing.

If it could improve assignment accuracy, it will be better