Fix quantization w/ DeepSpeed not working #32640
Closed
Conversation
This file contains 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. |
|
This fixed the issue with finetuning I reported here, thanks! |
SunMarc
approved these changes
Aug 13, 2024
Comment on lines
+441
to
+445
| if ( | ||
| not (hasattr(model, "hf_quantizer") and model.hf_quantizer.is_trainable) | ||
| and is_deepspeed_zero3_enabled() | ||
| and not getattr(model, "_transformers_zero3_init_used", True) | ||
| ): |
There was a problem hiding this comment.
What's the difference when the model is quantized vs no quantized ?
|
For my own understanding - why doesn't the user need to create the model after TrainingArguments in the quantized case but does in the unquantized case? |
|
Quantized weights aren't thrown into the |
|
Closing since we're doing a full revert |
5 tasks
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR tweaks the logic check introduced in #32299 to specifically exclude when Zero-3 is enabled and we're using a quantization method (which skips this chunk of the zero3 init use)
Fixes # (issue)
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@SunMarc can you check if what I've done here makes sense given the comment we're trying to fix?
#32299 (comment)
I believe so and I think the test makes sense, but you're quant eyes would be appreciated :)