Skip to content

Fix copy-paste error in phi4mm image processing #18315

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 1 commit into from
May 18, 2025
Merged

Fix copy-paste error in phi4mm image processing #18315

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 7 additions & 19 deletions vllm/model_executor/models/phi4mm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -415,15 +415,6 @@ class Phi4MMImagePixelInputs(TypedDict):
"""Shape: `(batch_size * num_images, H_mask, W_mask)`"""


class Phi4MMImageEmbeddingInputs(TypedDict):
type: Literal["image_embeds"]
data: Union[torch.Tensor, list[torch.Tensor]]
"""Shape: `(batch_size * num_images, image_feature_size, hidden_size)`

`hidden_size` must match the hidden size of language model backbone.
"""


class Phi4MMAudioFeatureInputs(TypedDict):
type: Literal["audio_features"]
data: Union[torch.Tensor, list[torch.Tensor]]
Expand All @@ -436,7 +427,6 @@ class Phi4MMAudioEmbeddingInputs(TypedDict):
"""Shape: `(batch_size, num_audios, audio_feature_size, hidden_size)"""


Phi4MMImageInput = Union[Phi4MMImagePixelInputs, Phi4MMImageEmbeddingInputs]
Phi4MMAudioInputs = Union[Phi4MMAudioFeatureInputs, Phi4MMAudioEmbeddingInputs]


Expand Down Expand Up @@ -1112,15 +1102,13 @@ def _parse_and_validate_multimodal_inputs(self, **kwargs: object) -> dict:

def _process_image_input(
self, image_input: Phi4MMImagePixelInputs) -> list[torch.Tensor]:
if image_input["type"] == "image_embeds":
image_embeds = image_input["image_embeds"].type(self.visual.dtype)
else:
dtype = next(self.vision_encoder.parameters()).dtype
pixel_values = image_input['data'].to(dtype)
image_sizes = image_input['image_sizes']
image_attention_mask = image_input['image_attention_mask']
image_embeds = self.vision_encoder(pixel_values, image_sizes,
image_attention_mask)

dtype = next(self.vision_encoder.parameters()).dtype
pixel_values = image_input['data'].to(dtype)
image_sizes = image_input['image_sizes']
image_attention_mask = image_input['image_attention_mask']
image_embeds = self.vision_encoder(pixel_values, image_sizes,
image_attention_mask)
return image_embeds

def get_multimodal_embeddings(
Expand Down