Skip to content

Add update_g_idx flag for setting qweight&g_idx #2143

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.

Sign up for GitHub

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

Merged
merged 8 commits into from
Apr 21, 2025
Merged

Add update_g_idx flag for setting qweight&g_idx #2143

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
2d8d920
add update g_idx flag
Kaihui-intel Mar 10, 2025
8c0cc57
add ipex_version match
Kaihui-intel Mar 18, 2025
d3b7220
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Mar 18, 2025
edc256d
Merge branch 'master' into kaihui/gidx_fix
XuehaoSun Mar 21, 2025
783fde6
replace ipex version match with var name
Kaihui-intel Apr 21, 2025
5c03050
revert ipex import
Kaihui-intel Apr 21, 2025
4714ad1
remove version import
Kaihui-intel Apr 21, 2025
529cb67
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Apr 21, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions neural_compressor/transformers/quantization/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -238,9 +238,16 @@ def _replace_linear(
dtype=torch.int32,
device=torch.device(device),
)

# Note: update_g_idx is only applicable for ipex versions >=2.7
model._modules[name].set_weights_bias(
module.qweight.data if hasattr(module, "qweight") else weight,
None if module.bias is None else module.bias.data,
**(
{"update_g_idx": not empty_weights}
if "update_g_idx" in model._modules[name].set_weights_bias.__code__.co_varnames
else {}
),
)
else:
raise Exception("{} device Unsupported weight only quantization!".format(device))
Expand Down