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

[Misc] Various cleanups for MM input processing (#29970)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-12-04 14:22:20 +08:00
committed by GitHub
parent 80f8af4b2f
commit 9ae2f60374
14 changed files with 67 additions and 225 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/model_executor/models/hunyuan_vision.py
View File

@@ -62,6 +62,7 @@ from vllm.multimodal.inputs import (
from vllm.multimodal.parse import (
DictEmbeddingItems,
ImageSize,
ModalityDataItems,
MultiModalDataItems,
MultiModalDataParser,
)
@@ -570,7 +571,7 @@ class HunYuanVLMultiModalDataParser(MultiModalDataParser):
def _parse_image_data(
self,
data: dict[str, torch.Tensor] | ModalityData[ImageItem],
):
) -> ModalityDataItems[Any, Any] | None:
if isinstance(data, dict):
return DictEmbeddingItems(
data,

4
vllm/model_executor/models/keye.py
View File

@@ -1000,7 +1000,7 @@ class KeyeMultiModalDataParser(MultiModalDataParser):
def _parse_image_data(
self,
data: dict[str, torch.Tensor] | ModalityData[ImageItem],
) -> ModalityDataItems[Any, Any]:
) -> ModalityDataItems[Any, Any] | None:
if isinstance(data, dict):
return DictEmbeddingItems(
data,
@@ -1017,7 +1017,7 @@ class KeyeMultiModalDataParser(MultiModalDataParser):
def _parse_video_data(
self,
data: dict[str, torch.Tensor] | ModalityData[VideoItem],
) -> ModalityDataItems[Any, Any]:
) -> ModalityDataItems[Any, Any] | None:
if isinstance(data, dict):
return DictEmbeddingItems(
data,

4
vllm/model_executor/models/keye_vl1_5.py
View File

@@ -333,7 +333,7 @@ class KeyeVL1_5MultiModalDataParser(MultiModalDataParser):
def _parse_image_data(
self,
data: dict[str, torch.Tensor] | ModalityData[ImageItem],
) -> ModalityDataItems[Any, Any]:
) -> ModalityDataItems[Any, Any] | None:
if isinstance(data, dict):
return DictEmbeddingItems(
data,
@@ -350,7 +350,7 @@ class KeyeVL1_5MultiModalDataParser(MultiModalDataParser):
def _parse_video_data(
self,
data: dict[str, torch.Tensor] | ModalityData[VideoItem],
) -> ModalityDataItems[Any, Any]:
) -> ModalityDataItems[Any, Any] | None:
if isinstance(data, dict):
return DictEmbeddingItems(
data,