How to keep it up?

[6801318d8d]

I don't understand how I'm supposed to study the book and at the same time keep it updated.

So I do:

git clone https://github.com/rasbt/LLMs-from-scratch
cd LLMs-from-scratch
git checkout -b my_branch

and work from there.

But then comes a new commit upstream.

That requires merging Jupyter Notebook code diffs into my branch.

Which is not easy to track down because it's JSON and it's not easy to realize what really changed.

May I suggest ditching the Jupyter Notebook file and upload only Python files generated with JupyText?

[casinca]

Hello, your work should be safe and free of merge conflicts. Every update Sebastian pushes is bonus content that is not in the book, it doesn't touch the .ipynb files linked to the book (unless a critical fix would be needed).
He expressly keeps the book content in the repo in sync/frozen with the physical copy to prevent divergence and avoid these problems for readers.

[rasbt]

Hi there,
The solution would be forking the repository first and then cloning the fork if you want to make your own updates. Otherwise, like @casinca said, the main book contents in the notebooks are not changed or updated. I am only adding bonus materials.

[6801318d8d]

The solution would be forking the repository first and then cloning the fork if you want to make your own updates.

I want to change the notebooks with my own stuff. I frequently make modifications to test things, print variables, change parameters, and in general to better understand what's going on.

But I also want the updates. Even if they are bonus content.

The problem is that if you push an update, I fetch from upstream, and merge there are going to be lots of merge conflicts in the JSON files of the python notebooks, which are not easy to solve.

My proposed solution would be for you to upload Python files, kept synchronized with the notebook through JupyText.

Otherwise, what is the best method to achieve this?