The use of the module `data_generation` in chapter 4

[NnannaOmoke]

Chapter 4, when defining Imports from data_generation.image_classification import generate_dataset

In the data_generation package as of 15/11/23, the sub_module image_classification no longer seems to exist. Are there any alternatives, so I may proceed with the chapter?

[dvgodoy]

Hi @NnannaOmoke

Thank you for your message. Are you running it on Google Colab? I just tried it out and it seems to be working there. The image_classification submodule that's imported can be found in the repo too:

https://github.com/dvgodoy/PyTorchStepByStep/blob/master/data_generation/image_classification.py

In Colab, these files should be downloaded by running the config_chapter4() function.

If you're running it locally, perhaps the file was removed. Please try either downloading it and placing it inside the data_generation folder where you're running the notebook, or, if you cloned the whole repo, you can also use git pull to retrieve the missing file directly from GitHub.

I hope this solves the issue.

Best, Daniel

[NnannaOmoke]

Thank you very much Mr. Godoy. I am actually running it (my notebooks) on VSCode, as I performed the setup of jupyter ,ipython and other packages manually. Cloning the repository and porting the folder data_generation to my working directory fixed the issue.

I must let you know that your book is perhaps the most clear and concise book I've come across on Deep Learning. As a computer science student pursuing a future in data science and ML, it has been a massive help. While I'm familiar with Tensorflow and Sci-kit Learn, will there be any books in the future coming out on those? Also, several concepts were mentioned to be out of the scope of this book. Are we going to get anything more in depth in the future too?

Again, thank you so very much for writing this book. I also really appreciate your timely reply Have a nice day, Mr. Godoy!

[dvgodoy]

Thank you for your kind words and the amazing feedback :-) I am happy to hear my book is so helpful to you!

I had the idea of writing a book focused on traditional ML and Scikit-Learn, but I'm not sure if this still makes sense in a post-ChatGPT world, to be honest. Won't be doing anything with TensorFlow, that's for sure.

Regarding concepts out of the scope, I had to draw the line somewhere because the book already had 1,000 pages :-) If you'd like some materials on ML, please check a repo I developed to teach a class a few years ago:

https://github.com/dvgodoy/ML_Fundamentals

Hopefully it will help with some concepts that weren't covered in the PyTorch book, such as train-validation split, etc.

Best, Daniel