biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[BUGFIX] Pixtral cannot be loaded with --limit-mm-per-prompt 0 (#33406)

Browse Source 
Signed-off-by: juliendenize <julien.denize@mistral.ai>
This commit is contained in:
Julien Denize
2026-01-30 11:52:02 +01:00
committed by GitHub
parent 10152d2194
commit 8e2ad97ad0
1 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
vllm/model_executor/models/pixtral.py
View File

@@ -70,7 +70,7 @@ from .interfaces import (
SupportsPP, SupportsPP,
) )
from .module_mapping import MultiModelKeys from .module_mapping import MultiModelKeys
from .utils import init_vllm_registered_model, maybe_prefix from .utils import StageMissingLayer, init_vllm_registered_model, maybe_prefix
from .vision import ( from .vision import (
VisionEncoderInfo, VisionEncoderInfo,
VisionFeatureSelectStrategy, VisionFeatureSelectStrategy,
@@ -93,6 +93,10 @@ except ImportError:
PATCH_MERGE = "patch_merge" PATCH_MERGE = "patch_merge"
def _is_layer_none_or_staged(layer: nn.Module) -> bool:
return layer is None or isinstance(layer, StageMissingLayer)
class PixtralImagePixelInputs(TensorSchema): class PixtralImagePixelInputs(TensorSchema):
""" """
Dimensions: Dimensions:
@@ -542,7 +546,7 @@ class PixtralForConditionalGeneration(
# Single pass over weights # Single pass over weights
for name, w in weights: for name, w in weights:
if is_vision_encoder_weights((name, w)): if is_vision_encoder_weights((name, w)):
if self.vision_encoder is None: if _is_layer_none_or_staged(self.vision_encoder):
continue continue
# Load vision encoder weights directly # Load vision encoder weights directly
trimmed_name = ".".join(name.split(".")[1:]) trimmed_name = ".".join(name.split(".")[1:])
@@ -551,7 +555,7 @@ class PixtralForConditionalGeneration(
with torch.no_grad(): with torch.no_grad():
default_weight_loader(param, w) default_weight_loader(param, w)
elif is_patch_merger((name, w)): elif is_patch_merger((name, w)):
if self.patch_merger is None: if _is_layer_none_or_staged(self.patch_merger):
continue continue
# Load vision patch merger weights directly # Load vision patch merger weights directly
trimmed_name = ".".join(name.split(".")[1:]) trimmed_name = ".".join(name.split(".")[1:])
@@ -559,7 +563,7 @@ class PixtralForConditionalGeneration(
with torch.no_grad(): with torch.no_grad():
default_weight_loader(param, w) default_weight_loader(param, w)
elif is_pre_mm_projector_norm((name, w)): elif is_pre_mm_projector_norm((name, w)):
if self.pre_mm_projector_norm is None: if _is_layer_none_or_staged(self.pre_mm_projector_norm):
continue continue
# Load vision pre_mm_projector_norm weights directly # Load vision pre_mm_projector_norm weights directly
trimmed_name = ".".join(name.split(".")[1:]) trimmed_name = ".".join(name.split(".")[1:])
@@ -567,7 +571,7 @@ class PixtralForConditionalGeneration(
with torch.no_grad(): with torch.no_grad():
default_weight_loader(param, w) default_weight_loader(param, w)
elif is_vision_lang_adapter_weights((name, w)): elif is_vision_lang_adapter_weights((name, w)):
if self.vision_language_adapter is None: if _is_layer_none_or_staged(self.vision_language_adapter):
continue continue
# Load vision-language adapter weights directly # Load vision-language adapter weights directly
trimmed_name = ".".join(name.split(".")[1:]) trimmed_name = ".".join(name.split(".")[1:])