[Bugifx][INC] Fix INC quantization method selection for non-quantized layers (#47237)
Signed-off-by: lvliang-intel <liang1.lv@intel.com> Co-authored-by: Kunshang Ji <kunshang.ji@intel.com> Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
This commit is contained in:
co-authored by
Kunshang Ji
mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
parent
754d5e1f65
commit
92ccd2c306
@@ -328,9 +328,11 @@ class INCConfig(QuantizationConfig):
|
||||
if (
|
||||
layer_name == prefix or layer_name == f"model.{prefix}"
|
||||
) and self.extra_config[layer_name].get("bits", 16) >= 16:
|
||||
if isinstance(layer, (LinearBase, ParallelLMHead)):
|
||||
return UnquantizedLinearMethod()
|
||||
if isinstance(layer, RoutedExperts):
|
||||
return UnquantizedFusedMoEMethod(layer.moe_config)
|
||||
return UnquantizedLinearMethod()
|
||||
return None
|
||||
|
||||
layer_config = self.config_parser.resolve(layer, prefix)
|
||||
if not layer_config.quantized:
|
||||
|
||||
Reference in New Issue
Block a user