Avis et commentaires pour d'étudiants pour Customising your models with TensorFlow 2 par Imperial College London

This is honestly one of the best courses I've ever done. I had completed the Tensorflow in Practise specialization by deeplearning.ai a couple of months back and took up this course as a refresher but this ended up being so much more! The lecture videos are top quality and explain the basics really well and the coding tutorial videos helped me get some much-needed practice. This course stays true to its name and covers important topics like designing custom models using the Model Subclassing API and using custom training loops. The assignments are very relevant to the course content and the capstone project, when finished, leaves you with a real sense of accomplishment and pride!

The lectures are clear and the coding assignments are very relevant and practical. The final project is complex but it is very rewarding once you complete it.

This is the best TF course in Coursera. The 4th week for Model subclassing and custom training loops is really informative and fantastic, which gives us full flexibility on TF model construction and training process.

Very well organized tour through Tensorflow 2 API, I learned a lot and enjoyed the course

great course

In short, take this course if you want a challenging course where you can learn TensorFlow 2 in depth.

I will add to my review on the first course of the specialization.Kevin and the GTAs do a brilliant job at mainting the assignments and autograders and the learning community is really helpful too in case you get stuck in some part of the assignment or the capstone project. Highly recommended!

This class is very good , I learned enough knowledge of tensorflow ,such as how to use tf.data ,how to embedding, how to do tokenlization, also I learned how to build customized tensorflow models

It just the last assignment of making a translation model, I had no idea where to start. It would be nicer to include a video to explaining the encoder and decoder mechanism

Excellent course! 100% recommended for anyone looking for more advanced TensorFlow knowledge.

Excellent course!

Overall, an exceptional and highly relevant course. I would have given it five stars, however some instructions on the capstone project were too vague, causing the project to take much more time to complete than is really necessary. Also, it appears that correctly completed neural network translators don't appear to produce very good translations, at least in the form we were asked to design, and I think there should be some comment about that from the instructor. Is it because the embeddings we were given were not that great? Was it because the network we designed was not deep enough to be effective, or was our custom training loop not well conceived? Thank you for developing and presenting this course. I especially appreciated Dr. Webster's clear and concise lecture videos. Overall, I thought the course nicely dovetailed with the two Andrew Ng courses I previously completed on Machine Learning and Deep Learning (with TensorFlow 1). This course helped me become a better programmer and was worth the effort I had to invest in it. Hopefully I will complete the final course in the specialization very soon, and launch my new career in AI software engineering!

I really like the course. It is repeated what I already knew but gave a lot of insight in customization. The high level course video are great they show the essence in a very clear and consice manner. I hope there are more courses like this coming. For me this was one of the best online courses I have done!!

The gap between the programming assignments and the cap stone projects is to wide

Best (and also hardest) coursera course I've completed so far. I particularly appreciate how the course let one get to grips with the TF documentation: when I started this course, that documentation was pretty opaque and incomprehensible to me, but now I find it a very valuable resource. The forums were great for clearing up problems, though sometimes I had to look very hard. The capstone project took me about twice as long as the suggested time, but then, my python skills ain't the best, as yet.

This course is very challenging, as require concrete understanding on tensorflow to conduct the whole project

Very useful course!!! Thanks!

Scope for improvement, for the RNN, LSTM, and Bi Directional layers.

Overall, not bad. But Capstone contains too much knowledge points that were taught in the previous labs and video lectures. A little stretch from the taught material is training and exercising, too much stretch is kind of waste of time. I took 4 weeks part timely to complete the Capstone project, which supposed to take an hour. And in general, I know my progress in other courses. So, I know this is out of the norm

Initially I wanted to do only Probabilistic DL (3rd course) because this material is not taught anywhere else as far as I am aware, but I learned quite a bit from other two courses as well even though I thought I knew the material. The entire specialisation is highly recommended, very good quality and very relevant content. The best of 2020 on Coursera, in my estimation.

Absolutely fantastic. The material is presented in a wonderfully concise and lucid way. The difficulty level also ramps up in a way that you really end up testing your understanding of what you've learned. Highly recommended.

I learned a lot from this course, thanks for providing this wonderful course. Can't wait to complete the last one, Probability with Tensorflow 2.

Loved this course, loved this specialization, the team doesn't support you so you are left alone. But we may see it as a formative experience.

Great Course, Got a lot to learn. Few things can be presented well especially in the 3rd the 4th week lab. Rest everything is good.

Amazing course, amazing specialization. I've always been afraid of Tensorflow, but thanks to this course that has totally changed.

I recumbent this course.

A lot of practice: notebooks, assessments, capstone project and just enough theory about TensorFlow