Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Bert training update #781

Merged
merged 5 commits into from
Feb 12, 2025
Merged

Bert training update #781

merged 5 commits into from
Feb 12, 2025

Conversation

tengomucho
Copy link
Collaborator

What does this PR do?

In this PR the BERT fine-tuning tutorial is updated to make it actually working and provide a reproducible result.
In the process, the text and script have been updated to reflect the necessary changes.
The notebook now will produce the documentation.
Finally, the old llama tutorial has been removed since redundant.

Before submitting

  • This PR improves the docs

@HuggingFaceDocBuilderDev

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Copy link
Collaborator

@dacorvo dacorvo left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, Thanks !

Copy link
Member

@michaelbenayoun michaelbenayoun left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I see that you generate the tutorial from the notebook. I think it is a great approach because it saves a lot of work. I have one small nit about it though: is it possible to not have the commands in the docs start by !? I know it's needed in the notebook to run bash commands, but I does not really make sense for the tutorial, wdty?

@tengomucho
Copy link
Collaborator Author

@michaelbenayoun I do not have a strong opinion towards/against the !. I see it is in some docs imported from notebooks(e.g.: here). I propose to leave it like that so if someone decides to run it on the notebook, it will work.

Since the tutorial is available in doc and notebook format, keep only
theh notebook, that can converted to md.
- Some links were wrong, they have been updated
- Some rewording was necessary
- The training script has been updated and simplified
- Transformers' Trainer was used in this example instead of
  NeuronTrainer, because otherwise an error occurs preventing to
  succeed.
- Output is updated accordingly.
This tutorial does pretty much the same as the SFT LoRA tutorial, so it
seems a bit redundant. It is hence removed.
@tengomucho
Copy link
Collaborator Author

force-pushed ot resolve conflicts.

@michaelbenayoun michaelbenayoun self-requested a review February 12, 2025 13:28
@tengomucho tengomucho merged commit 9c02cbd into main Feb 12, 2025
8 of 13 checks passed
@tengomucho tengomucho deleted the bert-training-update branch February 12, 2025 14:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants