vlm: support video as an input modality (#5888)

This commit is contained in:
Mick
2025-07-09 23:48:35 -07:00
committed by GitHub
parent 4ed57807c2
commit b5e3d6031c
42 changed files with 887 additions and 524 deletions
@@ -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,
)
]