14 nov. 2020
Excellent 'Introduction' to TensorFlow 2.0 (HINT: 'Introduction' does not mean 'Easy').
Evan Jones is at his best giving rapid intuitive explanations of advanced topics in deep neural networks.
17 mai 2020
I feel this course very valuable because it taught how to create an automated service in cloud with very huge data and working with distributed systems in production environment with minimal time.
par nicolas j•
8 déc. 2018
/!\ THIS COURSE IS A FRAUD /!\
this course is a way more about google cloud than tensorflow:
-You'll never learn how to install and setup tensorflow on your own infrastructure. Count how many time the logo of google cloud vs tensorflow appears. Another fact? they propose to win a google cloud tshirt on the forum... but you'll learn how to use google cloud...
-Tutors tell to don't use the forum on coursera, but the google tickets system... but as long as I pay on coursera, I expect to have a serious following from the tutors on the coursera forum, isn't it legit ?
-Of course they don't forget to send you email about google cloud products...
-You only have to pass quiz, no need to do the exercises, just log to them then your good to pass... is that serious ? what does worth my certificate ?
-I'm definitely not happy of this one, this will down my confidence in coursera...
btw I will send this message to coursera either in the way to get some explanation how this could be possible... I mean we are talking about google doing a fraud!
23 janv. 2019
even though the instructors present in a great way, especially the labs seem to be quite confusing and the videos couldn't prepare me so well for what expects me in the lab
par CHRISTOPHER M•
10 avr. 2019
This course needs to be revised for TensorFlow 2.0
par Chen C•
3 mars 2020
This course(especially Week 3 ) is not for Tensorflow beginners at all. The lecturer in Week 3 uses various concepts that have not been introduced at all. I wonder if I have already known so many concepts, why on earth I need to take this course. In week 3, a lab is given without any examples demonstrated in the video.
10 janv. 2019
While it is a fairly basic and informative course, I could was left disappointed with a few things
I found the lab infrastructure hard to setup and of limited use.
The lab exercise were trivial and not up to mark
The lab was not graded or no scope for us to run them independently on our cluster on our own cloud.
I was slightly irritated with trainers trying to promote google product instead of focusing on technical training.
Google BigQuery was an unnecessary addition and distracting till you realize you don't need it for this training.
par Ritayan G•
2 mai 2020
Goes from 0 to 100 really fast without any sort of teaching whatsoever. And within no time you are asked to write up an actual deployable ML model using core tensorflow. Which is just a tad difficult for beginners in Tensorflow.
par Andrew P B•
13 févr. 2021
It's an OK course, the examples are useful, but the labs often miss the balance of providing enough direction to guide the student, without telling the student what to do.
Content varies in quality and feels like it's patched together from a set of other offerings, meaning that there are occasional gaps (especially in the lectures) in the coverage.
As a course to complement other sources on ML and Tensorflow, it's OK, but I wouldn't take this course as your sole intro to the subjects covered.
par Edrian S•
8 août 2019
Older reviews mention that Datalab takes a really long time to spin up, and that the code is pre-written for you so that you don't really "learn". Not sure when things changed (although there is a tag that says material changed around the 1st of July 2019), but the labs no longer use Datalab. They use Jupyter Lab accessed via the AI Platform on GCP. It still takes a couple of minutes for Jupyter Lab to start up, but it's way faster than Datalab. They've also included two notebooks. One doesn't have much, if any, code written out asking you to try your hand at writing it. The other is the solution code. Anyway, wanted to make sure that update was reflected in the most current reviews. I completed the course on 8/6/2019.
par Sudesh A•
14 juil. 2018
The introduction to TensorFlow was good. Lab needs improvement; it would be helpful to have code templates that needs to be filled in by us to get credit for the lab, instead of just executing the code. Content in Estimator API module needs a bit more depth/explanation in my opinion.
par Eddie G•
15 janv. 2019
This course is not for those starting out with Machine Learning; its language is very technical and not learner-friendly. Furthermore, for the labs, a new log in is generated for each lab so that you need to setup the virtual machine new each time, which takes forever. Furthermore, the labs are not structured well as they don't accurately reflect what was taught in the same section.
par John D•
18 juil. 2018
Labs vms are to slow. Speaker is difficult to understand. Mic varies and speech pattern is not clear. The presentations need some graphics rather than a guy talking. Sketch out the ideas on a white board rather than talking 5 minutes to a single slide.
par ujwal s•
28 nov. 2019
Great discussion on TF infrastructure. Need to be updated to TF2.0
par Yaroslav L•
23 sept. 2021
Since TensorFlow is a core technology of Google AI and "Introduction to Tensorflow" is obvious choose for any person starting working in Google AI Platform, and therefore, probably most important Google Coursera cource, I expected that should be very carefully shaped course for beginners.
Cousre videos, readings a quizzles are fine, however, as it should be for programming course majority of ist is a programming labs and exersizes. And here we stepping into the troubles.
Labs of that course basically based on some tutorials where most of functional code cut off.
The way code cut from tutorials leave no chances, especially for beginners, to fill the gaps by themselves. For example, "Build sequencial model" #your code is here - without any guidance on what layers should be used, how much of them needed, which specific parameters need to set up. Even knowing what is possible to fill the gap, beginner unable to define from scratch which architecture is relevant for this case.
So there is only way to coplete notebooks - copy/paste.
Starting copypasting once it is hard to stop, and return to at least analysis of how this code work.
I think It would be more fair and usefull just provide link to tutorials. At least it skip useless copy-paste part of the work and concentrate on the code analysis.
When I say tutorials, I mean that some of the labs I traced back to original tutorials like https://www.tensorflow.org/tutorials/load_data/tfrecord
And for me it looks like that googlers made this course spent not so much time and efforts developing learning material specifically for this course.
par Bart V•
4 oct. 2020
Unfortunately, this is not a very good course.
There's a lot of talking, not supported by slides. The order of topics is unnatural. The teacher is mentioning all kinds of topics without explaining them.
The labs were quite frustrating, it's often not clear what is being asked.
I ended up following 2 courses by Imperial College London, also available on Coursera. These are really recommended over this one.
Then I returned to this one, but I have to maintain my judgement. This is not the course you need if you want to learn to use Tensorflow.
par Christopher W•
17 mars 2022
I've taken a lot of MOOCs, but this has to be one of the most frustrating ones, and this includes classes with unsupported/incompatible software requirements. The instruction is minimal, so if you're expected an introduction, good luck. Questions are asked out of order and out of context in the quizes, which you can take as many times as you want. The labs are arbitrary - all you have to do is open them. The actual lab tasks -if you choose to do them- are either ridiculously easy or are unclear and require familiarity with TensorFlow and the author's mind. This last bit is the most frustrating, because there's no way that I could have gotten through the labs without referencing the solution. I now know what a tensor is (week 1) and the difference between the sequential and functional models, but I would be hard pressed to build and deploy a model to GCP.
par Juan M P•
31 janv. 2019
Although the videos and content in general is OK, the environment and setup of Google DataLab for each lab is really disgusting – takes about 10 minutes to start with the proper exercise.
I understand that Google wants us to use their products, but the main purpose of this course (learning TensorFlow) is cluttered with this environment.
par Miika M•
21 nov. 2018
Don't think I'll remember much of what I've seen two weeks from now. Most of the time in the course was spent on spinning up the google cloud stuff. All the labs are done for you so no need to use your own brain. I'm very disappointed.
6 août 2018
labs were not properly working...
par Muhammad M M•
7 déc. 2020
Nothing is really properly explained, and labs expect you to be thorough with multiple topics, many of which are never covered. The readings are only vaguely related to the material required to solve the labs. The quizzes have many confusing questions which makes it seem like they were not proofread. And why do the labs have so much stuff required and a 45 minute timer?
par Thomas A•
4 juil. 2019
Course skips over a lot of important aspects for an introduction course. Doesn't properly test the items they teach. The labs are basically read the answer and try to figure it out because half the time it's not discussed in the videos.
par Brian R•
12 mars 2019
If you want videos rushing over 1 example of each high level element of a TF estimator and notebooks that people have put 0 effort into designing for actual learning and retention, this is the course for you.
par Ehsan F•
15 nov. 2018
nothing fancy or special or deep. just a superficial introduction.
22 juin 2021
Very good intro to Tensorflow 2.0.
The lab is really helpful to exercise Tensorflow APIs.
par Raghuram N•
29 avr. 2019
Good introductory course on Tensorflow.
par Aditya h•
11 août 2018
Pretty helpful in getting to know the various levels of abstractions of tensorflow API and avoiding various pitfalls while building the Tensorflow model