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

Fix: modelloader handling of model_kwargs load_in*bit #1999

Open
wants to merge 10 commits into
base: main
Choose a base branch
from

Conversation

NanoCode012
Copy link
Collaborator

Description

I missed one important case where model_kwargs["load_in_8bit"] and model_kwargs["load_in_4bit"] are deleted at the end of self.set_quantization_config() which would break all further dependency on the above kwarg. This PR fixes that and refactors the handling.

Motivation and Context

How has this been tested?

Screenshots (if appropriate)

Types of changes

Social Handles (Optional)

@NanoCode012 NanoCode012 changed the title Fix/model loader cast issues Fix: modelloader handling of model_kwargs load_in*bit Oct 28, 2024
@MengqingCao
Copy link
Contributor

This looks much better than checking if the key-value pair exists before each check. :-)

@NanoCode012
Copy link
Collaborator Author

To validate the new e2e test, I ran it on current main which threw an error.

main
image

this branch
image

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.

3 participants