docs: change assert to raise and some small docs

[pphuc25]

Hi,
I create a concise document outlining the process of improving code readability by replacing 'assert' statements with 'raise' statements, along with the addition of documentation and 'logger.info' statements.
I would like cc @stevhliu to review my PR.

[stevhliu]

Hi, in general we like to use raise statements instead of assert because the latter are considered more like debugging statements that can be toggled on or off in certain setups.

[pphuc25]

So to catch more debugging information, I switch from using assert to raise. I believe the code I am working on is correct, @stevhliu.

[stevhliu]

Sorry I read it the wrong way, you're correct! The rest of the changes look good to me. 🙂

Can you try running make style to fix the failing CI tests?

[pphuc25]

Hi @stevhliu, can you help me, I cannot run this command

[stevhliu]

Yes, it looks like you're missing black which can be fixed by installing it first:

pip install -e ".[quality]"

Then you should be able to run the make style command :)

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint.

[stevhliu]

Thanks @pphuc25! Pinging @LysandreJik for a final look :)

[LysandreJik]

Thank you for your changes @pphuc25, but let's focus on your changes from assert to raise statements for now.

Please revert the added comments/changes in logging statements; we want these scripts to be very minimal so we'll keep them the way they are currently for now.

Thank you!

[pphuc25]

Hi @LysandreJik, I have reverted them back. However, in run_mlm_no_trainer the logging do not have eval_loss, this seem the important one so user can know how well loss of their model is, I think we should keep add the eval_loss logging in that file.

[LysandreJik]

I have written down the changes that should be reverted in my opinion, the rest (including the eval loss), can be kept!

Thanks!

[LysandreJik]

Thank you for your PR!