Chevron Left
返回到 Data Pipelines with TensorFlow Data Services

學生對 deeplearning.ai 提供的 Data Pipelines with TensorFlow Data Services 的評價和反饋

4.4
454 個評分
92 條評論

課程概述

Bringing a machine learning model into the real world involves a lot more than just modeling. This Specialization will teach you how to navigate various deployment scenarios and use data more effectively to train your model. In this third course, you will: - Perform streamlined ETL tasks using TensorFlow Data Services - Load different datasets and custom feature vectors using TensorFlow Hub and TensorFlow Data Services APIs - Create and use pre-built pipelines for generating highly reproducible I/O pipelines for any dataset - Optimize data pipelines that become a bottleneck in the training process - Publish your own datasets to the TensorFlow Hub library and share standardized data with researchers and developers around the world This Specialization builds upon our TensorFlow in Practice Specialization. If you are new to TensorFlow, we recommend that you take the TensorFlow in Practice Specialization first. To develop a deeper, foundational understanding of how neural networks work, we recommend that you take the Deep Learning Specialization....

熱門審閱

PC

2020年4月16日

I understand why most of the students are furious about, but content wise, it one of those extremely helpful and important courses in Coursera. Really loved it!

GL

2020年3月2日

Laurence cares deeply about the students. Not only about what they learn, but that they actually enjoy and learn it. What a fantastic teacher.

篩選依據:

76 - Data Pipelines with TensorFlow Data Services 的 92 個評論(共 92 個)

創建者 Parth J

2020年6月3日

A lot of information was crammed in videos of very short duration. The course could have been more comprehensive. Also, a detailed explanation of errors on making wrong submissions would be very helpful.

創建者 András G

2020年2月17日

Dataset creation task was more complex for me then all previous before.

創建者 Shobhit G

2020年5月16日

Last Assignment does not have proper logs and instructions.

創建者 Tim.Ding

2020年9月15日

Very poor course experience due to assignment grader!!

創建者 Mark P

2020年4月27日

Assignments were very poor - especially week 4.

創建者 Jinxiang R

2020年3月1日

week 2 and week 4 is quite hard to follow

創建者 Sanket G

2020年9月22日

The final assignment is just annoying

創建者 Triantafyllos S

2020年10月17日

Fix the Week 4 Assignment please.

創建者 Cheuk L Y

2020年6月26日

That last week was awful.

創建者 David N

2021年3月20日

This course has excellent material, and as usual for all Coursera instructors that I have experienced so far (and for Laurence Moroney specifically) the instructor is extremely qualified, sincere, engaged, and tries to pack a ton of information into his/her course.

The reason I rated this course as a two star, was in hopes that it might stand out and the two suggestions below might get heard more (as I do very much think they would help). I hope that this rating won't hurt Laurences overall rating, as I assume he has thousands of high ratings to outweigh this one rating. If it does, please change the rating to 4 star.

Hear are my two suggestions:

(1) Point 1 -- Two many screen shots with just code! I believe Laurence is trying to pack as much info as he can into the course, and I do sincerely appreciate the detail and the specific code/how-tos/answers. But after a while, it is very close to experiencing "death by power-point". Especially, since we don't get copies of these lectures after our month subscription (that is a super bummer)! There is no way we can remember all this detail anyway (without working with this stuff), or we pause and write it all down. The super detailed notebooks are wonderful and repeat all this info anyway. I would much prefer to be able to keep the notebooks (if we can after class).

My recommendation/suggestion in this case would be to at least add more animation to code slides. In Andrew Ng's presentations, he walks you through all the steps, in many cases hand-writing out the math first. This would not be practical in this course, but there are cases, where Laurence high-lights the line of code he is talking about about. Much more of that is NEEDED. In fact it would be helpful to overlay arrows (as well) that show how parameters and arguments flow from one statement to the next in the code. You might think this is not necessary (as Laurence is very detailed in his explanations and that should be sufficient). But he does talk quickly, and it would be super helpful in not getting lost, and also in breaking up the eventual monotony that can set in with too many slides of just "code and talk".

Again the material is great, but the presentation is too much repetition of code and talk/explanation. If he were to tie it together better with more highlighted boxes an arrows connecting the flow (as Andrew Ng does in his presentations), I think that (or something similar) would help a lot.

(2) Point 2 -- You may disagree, but it I believe this is very important/helpful. I am not a fan of disease based presentations when learning. I know this subject is relevant to many apps, datasets, study and research today, but when you are trying to learn a new subject, it is fun (or helpful) to have to be looking at a bunch stats on diseases, which is a subject that can engender or call up fear in the audience. Subject matter that is neutral and not related to health problems is kinder to the learner/audience (imo). I would recommend avoiding lessons and topics on subjects that can engender or call up fear during the lesson. There are plenty other datasets (that you use) that are not focus on such domains/subjects.

創建者 Pavel

2020年3月11日

First three weeks were pretty interesting, especially pipelines and performance. However the examples and tasks were a little bit non-realistic. But Week 4 exercise was terrible. It is almost impossible to complete it using just grader's output. Running its copy in a separate colab's notebook is a must to be able to track errors, a lot of which are basically typos in names. The task description should have mentioned this much more explicitelly.

創建者 Fabrice L

2020年3月20日

That makes me sad to give such a bad rating, because I'm a big fan of Andrew Ng and DeepLearning.ai courses, but this one is really not at standart.

The lectures are confusing, we don't understand what's the goal of all that until week3.

The assignments can be a pain to pass, not because your code is wrong, but because you added a newline or modify a bit the cell.

And overall the topic is not very interesting, in an industry setting not useful.

創建者 Peer B

2020年10月19日

Normally the deeplearning.ai courses are really well done. This course unfortunately was an exception. The material was presented nicely, but the fourth week's programing exercise was a disaster. There was no way to complete it unless you reviewed several of the discussion forums to learn how to resolve the inherent bugs in the code and how to use the hidden week 5 notebook.

創建者 Daniel V H

2020年9月14日

homework assignments are not in the correct locations and assignments require significant digging to be able to submit and receive the appropriate grade. Please fix the issues so new students do not waste significant time. Moderators reading the discussion forums frequently would help.

創建者 Lim H H

2020年10月3日

The 4th week is terrible. The assignment was confusing and the lecture didn't prepare me to do it at all.

創建者 Mohannad B

2020年7月1日

It is a great course untell you watch the last week. Videos are very short and there is no easy way to debug the assignment. Wich was better.

創建者 Andy C

2020年8月27日

the week 4 exercise is so frustrating