A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 12, 06:00 UTC
[XPU] Fix AWQ skipped layer detection in IPEX quantization (#29774)
Signed-off-by: Fanli Lin <fanli.lin@intel.com>
Fanli Lin committed
Dec 1, 2025 at 20:00 UTC
f37e8938d26ad9c55d7bdef7098dfdba4204dd94
1 file changed
+4
-1
vllm/model_executor/layers/quantization/ipex_quant.py
+4
-1
@@ -150,7 +150,10 @@ class IPEXConfig(QuantizationConfig):
150
if isinstance(layer, LinearBase):
151
if self.method == "awq":
152
if is_layer_skipped(
153
- prefix, self.modules_to_not_convert, self.packed_modules_mapping
153
+ prefix,
154
+ self.modules_to_not_convert,
155
+ self.packed_modules_mapping,
156
+ skip_with_substr=True,
157
):
158
return UnquantizedLinearMethod()
159
return IPEXAWQLinearMethod(self)