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

[OV] Add --all-layers argument to CLI #713

Merged
merged 2 commits into from
May 17, 2024

Conversation

nikita-savelyevv
Copy link
Collaborator

What does this PR do?

Add --all-layers quantization argument to openvino export CLI interface.

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you make sure to update the documentation with your changes?
  • Did you write any new necessary tests?

@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.

action="store_true",
default=None,
help=(
"Whether embeddings and last MatMul layers should be compressed to a primary precision (usually, INT4)."
Copy link
Collaborator

Choose a reason for hiding this comment

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

Is there a non-INT4 usecase? I see below that if it is provided, all_layers is set to None if is_int8 else self.args.all_layers so it seems like it's ignored for INT8? If it is only for INT4 it would be good to clarify that in the help message. And possibly make it mutually exclusive with INT8.

Copy link
Collaborator

Choose a reason for hiding this comment

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

This is for INT4 only.

Copy link
Collaborator Author

@nikita-savelyevv nikita-savelyevv May 16, 2024

Choose a reason for hiding this comment

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

Thanks for your comment! By default we compress those layers to INT8, even if number of bits is set to 4. This flag allows to compress those layer to INT4 as well. I've updated the description, hopefully it is more clear now (since currently the only supported primary precision is INT4, I've rephrased it a bit).

Copy link
Collaborator

@echarlaix echarlaix 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 @nikita-savelyevv

@echarlaix echarlaix merged commit bc5051f into huggingface:main May 17, 2024
11 checks passed
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.

5 participants