Suggestions for Decouple gradient accumulation from the number of minibatches generated #3396
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
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. |
edbeeching
approved these changes
May 2, 2025
edbeeching
approved these changes
May 2, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Some suggestions for PR #3388:
generation_batch_sizeunder the sectionParameters that control generationint | Noneinstead ofint | strbecause the the CLI parser doesn't support multi-type (that's why the CI is failing)per_device_train_batch_sizeper_device_train_batch_size * gradient_accumulation_stepsper_device_train_batch_size * world_sizeper_device_train_batch_size * world_size * gradient_accumulation_stepsper_device_train_batch_size * steps_per_generationper_device_train_batch_size * world_size * steps_per_generationnum_generationswill make less sense to use with this PR, I removed it from the figure, and hopefully gained clarityself.args._steps_per_generation < self.args.gradient_accumulation_steps(not sure why you'd want that, but it just requires to change!=into<.