educative.io

Educative

"Deep learning with TensorFlow" needs to be rewritten

I 'm just submitting this to give some feedback. I am going through “machine learning for software engineers” module, which was amazing and very well thought until I reached this section. It is so badly written I am having so much trouble following it. The order doesn’t make any sense. The overall code that I am supposed to write has many holes, so that If I want to reproduce the steps in a project of mine, I will end up following some other tutorial to figure it out. I’m telling you, having seen educative’s amazing work suck as the “grokking the system design interview”, this is way below the platform’s level. Please revisit.


Course: Educative: Interactive Courses for Software Developers
Lesson: Educative: Interactive Courses for Software Developers

2 Likes

hi @Vaggelis_Petratos,
Thank you very much for your honest feedback; we much appreciate it. And we will undoubtedly review and rework these lessons. Thanks.

I second the feedback by @Vaggelis_Petratos. Definitely a departure from the earlier sections, so much so, that it creates an impression this section, in style and substance, was written by a different author. I am sure an NLP classifier may agree :wink:.

1 Like

A related issue:

In the Time to Code section: Time to Code
I suspect there is an error in the “for loop”. Normally, an array (or dictionary of arrays, in this case, is generated from the individual inputs prior to executing a session.run method. However, the offered solution executes the run method internal to the loop. Assuming this is the correct solution, the code will create 1000 instances of the session object. Is this the desired result?

Offered Solution:
# input_data, input_labels, inputs, labels, train_op
# are all predefined in the backend
for i in range(1000):
# CODE HERE
feed_dict = {
inputs: input_data[i],
labels: input_labels[i]
}
sess.run(train_op,feed_dict=feed_dict)

Proposed Solution:
# input_data, input_labels, inputs, labels, train_op
# are all predefined in the backend
for i in range(1000):
# CODE HERE
feed_dict = {
inputs: input_data[i],
labels: input_labels[i]
}
sess.run(train_op,feed_dict=feed_dict)

Hi @Leo_Pimentel

Yes, this is the correct and desired output. For reference, I have attached the screenshot here.

I hope that this guide is helpful. Remember that I am always available via message to help you with any difficulty you might encounter.

Regards,

Happy Learning :slight_smile:

1 Like

I am writing this 8 months after Vaggelis’s comment. The course “Deep Learning with TensorFlow” still suffers the same issues noted. The code written does not follow a good organization, in the sense that by following the lessons you won’t come up with a code base that works and your thoughts has a lot of unknowns as of how a working code should be written.


Course: https://www.educative.io/courses/ml-for-software-engineers
Lesson: Educative: Interactive Courses for Software Developers