[V1] Enable multi-input by default (#15799)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-04-12 08:52:39 +00:00
committed by GitHub
parent f069f3ea74
commit d9fc8cd9da
21 changed files with 214 additions and 105 deletions
+1 -1
View File
@@ -126,7 +126,7 @@ class MiniCPMOMultiModalDataParser(MiniCPMVMultiModalDataParser):
def _parse_audio_data(
self,
data: Union[dict[str, torch.Tensor], ModalityData[AudioItem]],
) -> ModalityDataItems[Any, Any]:
) -> Optional[ModalityDataItems[Any, Any]]:
if isinstance(data, dict):
return MiniCPMOAudioEmbeddingItems(
data,
+2 -2
View File
@@ -290,7 +290,7 @@ class MiniCPMVMultiModalDataParser(MultiModalDataParser):
def _parse_image_data(
self,
data: Union[dict[str, torch.Tensor], ModalityData[ImageItem]],
) -> ModalityDataItems[Any, Any]:
) -> Optional[ModalityDataItems[Any, Any]]:
if isinstance(data, dict):
return MiniCPMVImageEmbeddingItems(
data,
@@ -302,7 +302,7 @@ class MiniCPMVMultiModalDataParser(MultiModalDataParser):
def _parse_video_data(
self,
data: Union[dict[str, torch.Tensor], ModalityData[VideoItem]],
) -> ModalityDataItems[Any, Any]:
) -> Optional[ModalityDataItems[Any, Any]]:
if isinstance(data, dict):
return MiniCPMVVideoEmbeddingItems(
data,
+2 -2
View File
@@ -720,7 +720,7 @@ class Qwen2VLMultiModalDataParser(MultiModalDataParser):
def _parse_image_data(
self,
data: Union[dict[str, torch.Tensor], ModalityData[ImageItem]],
) -> ModalityDataItems[Any, Any]:
) -> Optional[ModalityDataItems[Any, Any]]:
if isinstance(data, dict):
return DictEmbeddingItems(
data,
@@ -734,7 +734,7 @@ class Qwen2VLMultiModalDataParser(MultiModalDataParser):
def _parse_video_data(
self,
data: Union[dict[str, torch.Tensor], ModalityData[VideoItem]],
) -> ModalityDataItems[Any, Any]:
) -> Optional[ModalityDataItems[Any, Any]]:
if isinstance(data, dict):
return DictEmbeddingItems(
data,