-
Notifications
You must be signed in to change notification settings - Fork 27.4k
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
Update no_trainer scripts with new Accelerate functionalities #16617
Conversation
The documentation is not available anymore as the PR was closed or merged. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for working on those examples! Left a couple of comments, but looks pretty good already!
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for duplicating the effort across all the examples!
Update the
no_trainer
scripts to keep aligned with Accelerate capabilitiesWhat does this add?
Updates all
no_trainer
scripts to use the latest capabilities.Why is it needed?
Accelerate had a number of new capabilities added, including better saving/loading, experiment tracking, and support for LR Schedulers. As a result, much of the current scripts can either be simplified from their hard-coded behaviors, or have these features added
Modified scripts with potential major changes:
language-modeling
multiple-choice
question-answering
summarization
text-classification
token-classification
translation
The speech fine-tuning will be updated in a later PR
Basic usage examples:
accelerate launch language-modeling/run_clm_no_trainer --checkpointing_steps "epoch"
accelerate launch language-modeling/run_clm_no_trainer --resume_from_checkpoint "epoch_1"
Anticipated maintence burden? (What will happen in say, 3 months if something changes)
As it gets more widly used, these scripts might need small updates if we find the end-users prefer a different experience when it comes to logging, or other small bugfixes we find as time goes on.