Skip to content

Migrate InternVLImagePixelInputs (in nemotron_vl.py) to TensorSchema #22023

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Aug 19, 2025
Merged

Migrate InternVLImagePixelInputs (in nemotron_vl.py) to TensorSchema #22023

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b2d5615
Migrate InternVLImagePixelInputs (in nemotron_vl.py) to TensorSchema
bbeckca Jul 31, 2025
5133a18
Merge branch 'main' into nemotron-vl
DarkLight1337 Aug 18, 2025
e1fa6fe
Merge branch 'main' into nemotron-vl
DarkLight1337 Aug 19, 2025
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
28 changes: 5 additions & 23 deletions vllm/model_executor/models/nemotron_vl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -458,27 +458,6 @@ def extract_feature(self, pixel_values: torch.Tensor) -> torch.Tensor:
vit_embeds = self.mlp1(vit_embeds)
return vit_embeds

def _validate_pixel_values(self, data: torch.Tensor) -> torch.Tensor:

#use force_image_size to get image_size
h = w = self.config.force_image_size
expected_dims = (3, h, w)

def _validate_shape(d: torch.Tensor):
actual_dims = tuple(d.shape)

if actual_dims != expected_dims:
expected_expr = str(expected_dims)
raise ValueError(
"The expected shape of pixel values per image per batch "
f" per patch is {expected_expr}. "
f"You supplied {tuple(d.shape)}.")

for d in data:
_validate_shape(d)

return data

def _parse_and_validate_image_input(
self, **kwargs: object) -> Optional[InternVLImageInputs]:
pixel_values_flat = kwargs.pop("pixel_values_flat", None)
Expand Down Expand Up @@ -516,9 +495,12 @@ def _parse_and_validate_image_input(

return InternVLImagePixelInputs(
type="pixel_values",
pixel_values_flat=self._validate_pixel_values(
pixel_values_flat),
pixel_values_flat=pixel_values_flat,
num_patches=image_num_patches,
resolve_bindings={
"h": self.config.force_image_size,
"w": self.config.force_image_size
},
)

raise AssertionError("This line should be unreachable.")
Expand Down