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

[Misc] Standardize handling of mm_processor_kwargs.size (#35284)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-02-26 21:05:46 +08:00
committed by GitHub
parent ec13e549d3
commit 845ee348ef
9 changed files with 135 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
vllm/model_executor/models/ernie45_vl.py
View File

@@ -829,16 +829,31 @@ class Ernie4_5_VLProcessingInfo(BaseProcessingInfo):
spatial_conv_size = hf_config.spatial_conv_size
temporal_conv_size = hf_config.temporal_conv_size
if self.ctx.model_config.trust_remote_code:
# Defined in HF Hub repo
min_pixels_key = "min_pixels"
max_pixels_key = "max_pixels"
else:
# Defined in Transformers library (requires v5.0 or above)
min_pixels_key = "shortest_edge"
max_pixels_key = "longest_edge"
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {min_pixels_key: override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {max_pixels_key: override_max_pixels}
if do_resize:
resized_height, resized_width = smart_resize(
height=image_height,
width=image_width,
factor=patch_size * spatial_conv_size,
min_pixels=size["min_pixels"],
max_pixels=size["max_pixels"],
min_pixels=size[min_pixels_key],
max_pixels=size[max_pixels_key],
)
preprocessed_size = ImageSize(width=resized_width, height=resized_height)
else:

8
vllm/model_executor/models/hunyuan_vision.py
View File

@@ -636,7 +636,13 @@ class HunYuanVLProcessingInfo(BaseProcessingInfo):
spatial_merge_size = vision_config.spatial_merge_size
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {"shortest_edge": override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {"longest_edge": override_max_pixels}
if do_resize:
resized_height, resized_width = smart_resize(

8
vllm/model_executor/models/keye.py
View File

@@ -1021,7 +1021,13 @@ class KeyeProcessingInfo(BaseProcessingInfo):
temporal_patch_size = 1
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {"min_pixels": override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {"max_pixels": override_max_pixels}
if do_resize:
resized_height, resized_width = smart_resize(

21
vllm/model_executor/models/paddleocr_vl.py
View File

@@ -155,15 +155,30 @@ class PaddleOCRVLProcessingInfo(BaseProcessingInfo):
patch_size = vision_config.patch_size
merge_size = vision_config.spatial_merge_size
if self.ctx.model_config.trust_remote_code:
# Defined in HF Hub repo
min_pixels_key = "min_pixels"
max_pixels_key = "max_pixels"
else:
# Defined in Transformers library (requires v5.0 or above)
min_pixels_key = "shortest_edge"
max_pixels_key = "longest_edge"
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {min_pixels_key: override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {max_pixels_key: override_max_pixels}
resized_height, resized_width = smart_resize(
height=image_height,
width=image_width,
factor=patch_size * merge_size,
min_pixels=size["min_pixels"],
max_pixels=size["max_pixels"],
min_pixels=size[min_pixels_key],
max_pixels=size[max_pixels_key],
)
preprocessed_size = ImageSize(width=resized_width, height=resized_height)

17
vllm/model_executor/models/qwen2_vl.py
View File

@@ -843,7 +843,13 @@ class Qwen2VLProcessingInfo(BaseProcessingInfo):
temporal_patch_size = vision_config.temporal_patch_size
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {"shortest_edge": override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {"longest_edge": override_max_pixels}
if do_resize:
resized_height, resized_width = smart_resize(
@@ -930,7 +936,14 @@ class Qwen2VLProcessingInfo(BaseProcessingInfo):
image_processor = self.get_image_processor()
mm_kwargs = self.ctx.get_merged_mm_kwargs({})
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {"shortest_edge": override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {"longest_edge": override_max_pixels}
max_pixels = size["longest_edge"]
unit = patch_size * merge_size

8
vllm/model_executor/models/qwen3_vl.py
View File

@@ -647,7 +647,13 @@ class Qwen3VLProcessingInfo(Qwen2VLProcessingInfo):
temporal_patch_size = vision_config.temporal_patch_size
mm_kwargs = self.ctx.get_merged_mm_kwargs(mm_kwargs)
size = mm_kwargs.get("size", image_processor.size)
size = image_processor.size
if override_size := mm_kwargs.get("size"):
size = size | override_size
if (override_min_pixels := mm_kwargs.get("min_pixels")) is not None:
size = size | {"shortest_edge": override_min_pixels}
if (override_max_pixels := mm_kwargs.get("max_pixels")) is not None:
size = size | {"longest_edge": override_max_pixels}
if do_resize:
if is_video: