Skip to content

reduce AutoGPTQ priority and add transformer requirement#1467

Merged
xin3he merged 2 commits intomainfrom
xinhe/2-25
Feb 28, 2026
Merged

reduce AutoGPTQ priority and add transformer requirement#1467
xin3he merged 2 commits intomainfrom
xinhe/2-25

Conversation

@xin3he
Copy link
Copy Markdown
Contributor

@xin3he xin3he commented Feb 25, 2026

Description

reduce AutoGPTQ priority and add transformer requirement

Type of Change

  • Bug fix
  • New feature
  • Documentation update
  • Performance improvement
  • Code refactoring
  • Other (please specify):

Related Issues

Fixes or relates to #

Checklist Before Submitting

  • My code has been tested locally.
  • Documentation has been updated as needed.
  • New or updated tests are included where applicable.

Copilot AI review requested due to automatic review settings February 25, 2026 09:32
Signed-off-by: He, Xin3 <xin3.he@intel.com>
Copy link
Copy Markdown
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

This PR adjusts backend selection for AutoGPTQ inference by lowering its selection priority and explicitly constraining it to Transformers versions it supports, so that more actively maintained backends are preferred while still allowing AutoGPTQ usage on compatible environments.

Changes:

  • Removed the runtime gating that only registered AutoGPTQ backends when transformers<5.
  • Lowered auto_gptq:exllamav2 backend priority so it’s less likely to be auto-selected over other GPTQ-capable backends.
  • Added explicit transformers<5.0.0 requirement to all AutoGPTQ backends to enforce compatibility.

Signed-off-by: He, Xin3 <xin3.he@intel.com>
@xin3he xin3he merged commit e7e1c78 into main Feb 28, 2026
29 checks passed
@xin3he xin3he deleted the xinhe/2-25 branch February 28, 2026 05:52
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