Skip to content

Conversation

violetch24
Copy link
Contributor

@violetch24 violetch24 commented Feb 26, 2024

Type of Change

Migrate SmoothQuant for IPEX to 3.x API, API changed

Description

To-do:

  • Enable different scales for Linears sharing the same input
  • Add more test cases (autotune)
  • Add accuracy test

Expected Behavior & Potential Risk

How has this PR been tested?

UT passed

Dependency Change?

@violetch24 violetch24 marked this pull request as ready for review February 26, 2024 08:54
@violetch24 violetch24 requested review from xin3he and yiliu30 February 26, 2024 08:54
Copy link
Contributor

@yiliu30 yiliu30 left a comment

Choose a reason for hiding this comment

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

Left some comments, others are LGTM.

violetch24 and others added 13 commits February 26, 2024 17:56
Signed-off-by: Cheng, Zixuan <[email protected]>
Signed-off-by: Cheng, Zixuan <[email protected]>
* fix conflict

Signed-off-by: Lu, Yintong <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Lu, Yintong <[email protected]>
Co-authored-by: Lu, Yintong <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Signed-off-by: Cheng, Zixuan <[email protected]>
Signed-off-by: Cheng, Zixuan <[email protected]>
Signed-off-by: Cheng, Zixuan <[email protected]>
@chensuyue chensuyue added this to the v2.5 milestone Feb 28, 2024
@xin3he
Copy link
Contributor

xin3he commented Feb 29, 2024

@violetch24 Unittest is not enough, please migrate UTs pure for algorithms and more sq accuracy check.

@xin3he
Copy link
Contributor

xin3he commented Feb 29, 2024

Currently, we have enabled save & load API. Please refer to fp8 UT's test_save_load

Copy link
Contributor

@xin3he xin3he left a comment

Choose a reason for hiding this comment

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

There will be another PR to polish SmoothQuant UTs.

@chensuyue chensuyue merged commit a5e5f5f into master Feb 29, 2024
@chensuyue chensuyue deleted the zixuan/sq_3x branch February 29, 2024 09:03
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