vlm: support video as an input modality (#5888)
This commit is contained in:
@@ -106,7 +106,7 @@ class PixtralProcessor(BaseMultimodalProcessor):
|
||||
pixel_values=processor_output["pixel_values"],
|
||||
image_sizes=processor_output["image_sizes"],
|
||||
modality=Modality.IMAGE,
|
||||
image_offsets=image_offsets,
|
||||
offsets=image_offsets,
|
||||
)
|
||||
]
|
||||
|
||||
|
||||
Reference in New Issue
Block a user