I would like to thank Lak and Chris for their wonderful presentation of the deployment of ML models on the Google Cloud Platform. The case study problem chosen for the course is also unique.
awesome learning experience fro the teacher from google. thanks to coursera and google for providing me such a good lesson which will be beneficial for my upcoming future and research work
By Karl J•
It has great content for the lectures and a great introduction to GCP to make a production machine learning service. The labs are helpful to an extent, but you're on your own to troubleshoot or explore for labs because there is no help on discussion boards. They say to email tech support, but tech support's reponse was to do the lab as written, despite being told that's what was done. It's hard to get through a lab searching StackOverflow or Google because of the time limits.
By David K•
Good: Course structure = great, content is relevant and interesting
Bad: Labs do not always work (e.g. deprecated GCP modules incompatible with apache-beam), code for labs already contains answers... would be nice to have "lab" file and "answer" file to make learning more explicit, also, the white guy with the mustache should rerecord his videos.... the cadence is distracting and he does not go into as much depth as Lak
By Mustafa H K•
- the lectures are not easy to follow sometimes, might be better if the code is shown more step by step rather than being shown all at once and then the instructor arbitrarily starts explaining from a particular location.
the labs are either too easy or such that there isn't enough information to find out the answers unless you really do arbitrary try and error. Either way you don't come out learning much.
By Mateo V•
Lectures are good but the labs are just tapping shift +enter on jupyter notebooks (There is no really hands-on). Some intermediate steps on notebooks are not well explained. For learning is ok , but having to pay for a certificate on this , I wouldn't recommend it
By Ian Q T C•
Exactly what it says. Labs are trivial and I felt like I didn't learn much other than how to use the interface for serving and taking a model from start to finish. The core concepts are useful both in GCP and if you decide to roll your own stack
Good to work in a real world environment, though not much hands-on work (most code has already been provided). Still not familiar with API's details.
This course is not very practical nor very well explained. The topic is very interesting but it is not delivered clear enough.
By Eun S J•
I am satisfied with GCP training except for some errors.
I think I need the latest update.
By Jitesh B C•
more explanation about labs were needed for help to understand the code
the course is helpful for any learner initial to touch GCP learning
By Harshit S•
Deployment was not successful. Was not able to debug
By AMAN V•
The labs could have been a bit more useful.
By Bruno N H P•
Some codes of yours notebooks doesnt work
By Marian P•
Some labs have to be reviewed and updated
By Rahul D G•
QuickLabs has error in many labs
By Mark Y•
By Cody G•
Really, really poor due to the fact that it's outdated to the point being broken in various ways. For instance, Qwiklabs confirms there is a bug in the local prototyping lab that prevents us from actually training anything. There is no consensus between video and lab as to whether we are using Keras or Estimators. Lab 5 points to an old ipynb using Estimators, while Lab 6 assumes we have used a newer version featuring Keras--so one redoes all of Lab 5. There is a "Optional Lab" with accompanying videos, but where there is no option to do the lab--the lab has been removed for half a year or more (per the discussion board). The video summary mentions that we have used App Engine to make a frontend for our model--no we didn't! I understand having a few bugs, but this course surely wasted several hours of my time. And yes, as others mention, the TODO's, for the most part, don't require much thought. On the upside, I think Lak is a really excellent presenter, and that Chris and the other guy are also pretty good.
By Carl S•
The labs were out of sync with the reviews and the code was either 100% complete or 0% complete, there wasn't much of an opportunity to learn.
Across the 1 course they used 2 different notebook applications as it was updated, which was confusing. They also scattered Tensorflow / Tensorflow.Keras throughout and it lacked consistency.
Also the demonstrators shouldn't speak into the mic so closely.
By Nikhileshkumar I•
Course content is good. But Coursera policy is bad. After completion of the course, only a 1-month extension is given to revise it after that. You cant see the content. This is very bad. It would be better to download the videos.
How are you supposed to revise the contents?
After that, you have to pay again. I will think twice before going for any course with coursera.
By Luis A S•
Some content is out-of-date and the labs are mostly solved with no much guidance. I was expecting a better explanation on what are the key points in the labs, such as the commands needed to create training jobs and so on.
By Han L•
Good content, but flawed lab design. The labs are designed to run in sequence and have dependencies from previous labs. Those dependencies break with different sessions of Qwiklabs.
By Cathy W•
Videos were helpful to get to know Google Cloud services, but the labs were a mess with different versions (deepdive vs deepdive2) that were not a pleasant experience.