fix(quantization): Fix AWQ dequantize on Intel XPU and refactor AutoAWQ config (#42727)

Signed-off-by: Alex <alex.tech.lab@outlook.com>
Signed-off-by: AlexHuang <jihuihuang@tencent.com>
Co-authored-by: Claude <noreply@anthropic.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:
AlexHuang
2026-06-18 20:12:28 +08:00
committed by GitHub
co-authored by Claude Kunshang Ji mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
parent 351c72d6e5
commit 8d4f54966c
20 changed files with 580 additions and 429 deletions
+2 -2
View File
@@ -26,7 +26,7 @@ from vllm.model_executor.layers.linear import (
RowParallelLinear,
)
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.auto_awq import AutoAWQConfig
from vllm.multimodal import MULTIMODAL_REGISTRY
from vllm.multimodal.inputs import (
MultiModalFieldConfig,
@@ -420,7 +420,7 @@ class Cohere2VisionForConditionalGeneration(
):
# the awq models from OpenGVLab missing `modules_to_not_convert`
# patch the quant_config to add `modules_to_not_convert` back
if isinstance(quant_config, AWQConfig):
if isinstance(quant_config, AutoAWQConfig):
text_config = config.text_config
llm_quant_config = getattr(text_config, "quantization_config", None)
if (not quant_config.modules_to_not_convert) and (
+2 -2
View File
@@ -20,7 +20,7 @@ from vllm.config import VllmConfig
from vllm.config.multimodal import BaseDummyOptions
from vllm.inputs import MultiModalDataDict
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.auto_awq import AutoAWQConfig
from vllm.model_executor.models.intern_vit import (
InternVisionModel,
)
@@ -608,7 +608,7 @@ class InternVLChatModel(
):
# the awq models from OpenGVLab missing `modules_to_not_convert`
# patch the quant_config to add `modules_to_not_convert` back
if isinstance(quant_config, AWQConfig):
if isinstance(quant_config, AutoAWQConfig):
text_config = config.text_config
llm_quant_config = getattr(text_config, "quantization_config", None)
if (not quant_config.modules_to_not_convert) and (
+2 -2
View File
@@ -11,7 +11,7 @@ from vllm.config import VllmConfig
from vllm.model_executor.layers.linear import ReplicatedLinear
from vllm.model_executor.layers.pooler import DispatchPooler
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.auto_awq import AutoAWQConfig
from vllm.model_executor.models.internvl import (
BaseInternVLDummyInputsBuilder,
BaseInternVLMultiModalProcessor,
@@ -144,7 +144,7 @@ class LlamaNemotronVLChatModel(nn.Module, SupportsMultiModal, SupportsPP, Suppor
):
# the awq models from OpenGVLab missing `modules_to_not_convert`
# patch the quant_config to add `modules_to_not_convert` back
if isinstance(quant_config, AWQConfig):
if isinstance(quant_config, AutoAWQConfig):
text_config = config.get_text_config()
llm_quant_config = getattr(text_config, "quantization_config", None)
if (not quant_config.modules_to_not_convert) and (
+2 -2
View File
@@ -19,7 +19,7 @@ from vllm.config.multimodal import BaseDummyOptions
from vllm.inputs import MultiModalDataDict
from vllm.model_executor.layers.linear import ReplicatedLinear
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.quantization.awq import AWQConfig
from vllm.model_executor.layers.quantization.auto_awq import AutoAWQConfig
from vllm.model_executor.models.intern_vit import (
InternVisionModel,
)
@@ -205,7 +205,7 @@ class SkyworkR1VChatModel(nn.Module, SupportsMultiModal, SupportsPP):
):
# the awq models from OpenGVLab missing `modules_to_not_convert`
# patch the quant_config to add `modules_to_not_convert` back
if isinstance(quant_config, AWQConfig):
if isinstance(quant_config, AutoAWQConfig):
text_config = config.text_config
llm_quant_config = getattr(text_config, "quantization_config", None)
if (not quant_config.modules_to_not_convert) and (