Skip to content

[Bugfix] Fix getting vision features in Transformer Multimodal backend #32933

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
DarkLight1337 merged 2 commits into from
Jan 23, 2026
+9 −0
Merged

[Bugfix] Fix getting vision features in Transformer Multimodal backend #32933

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
333a270
get_image_feats
zucchini-nlp Jan 23, 2026
93dbc7c
can be dict by default sometimes
zucchini-nlp Jan 23, 2026
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
9 changes: 9 additions & 0 deletions vllm/model_executor/models/transformers/multimodal.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -376,6 +376,15 @@ def embed_multimodal(self, **kwargs):
pixel_values, **kwargs
)

# Transformers `v5`, `self.get_image_features` returns a tuple
# containing the features and optionally attentions/hidden_states
# After v5 is settled, we can enable qwen3-vl with several outputs
# from `self.get_image_features`
if isinstance(vision_embeddings, tuple):
vision_embeddings = vision_embeddings[0]
elif isinstance(vision_embeddings, dict):
vision_embeddings = vision_embeddings.pooler_output
Comment thread
zucchini-nlp marked this conversation as resolved.

if isinstance(vision_embeddings, torch.Tensor):
if vision_embeddings.ndim == 2:
vision_embeddings = vision_embeddings.unsqueeze(0)
Expand Down