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

PyTorch Weight-only 3x API Implementation #1549

Closed
wants to merge 3 commits into from
Closed

PyTorch Weight-only 3x API Implementation #1549

wants to merge 3 commits into from

Conversation

xin3he
Copy link
Contributor

@xin3he xin3he commented Jan 18, 2024

Type of Change

Migration weight-only implementation to 3x API.

Description

  • Design review
  • Refine weight-only configuration
    • RTN
    • GPTQ
    • AWQ
    • TEQ
  • Separate public utilities from weight_only algos
    • RTN
    • GPTQ
    • AWQ
    • TEQ
  • Unify interface for quantization
    • RTN
    • GPTQ
    • AWQ
    • TEQ
  • Migrate algos.
    • AWQ
    • TEQ
  • Implement double_quant.
    • AWQ
    • TEQ
  • Add UTs for each algos on global configuration.
    • RTN
    • GPTQ
    • AWQ
    • TEQ
  • Add UTs for each algos on per-op configuration quantization.
    • RTN
    • GPTQ
    • AWQ
    • TEQ

Expected Behavior & Potential Risk

UT pass

@xin3he xin3he added the draft label Jan 18, 2024
@xin3he xin3he changed the title Xinhe/woq PyTorch Weight-only 3x API Implementation Jan 18, 2024
@xin3he xin3he marked this pull request as draft January 18, 2024 06:04
@xin3he
Copy link
Contributor Author

xin3he commented Jan 18, 2024

For INC unification, PR: Rename RTNWeightOnlyConfig to RTNConfig merge first.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
draft PyTorch Related to PyTorch F/W WIP
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants