Skip to content
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

Multimodal placeholder WA #734

Draft
wants to merge 1 commit into
base: habana_main
Choose a base branch
from
Draft

Multimodal placeholder WA #734

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
10 changes: 10 additions & 0 deletions vllm/model_executor/models/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -360,6 +360,16 @@
Note:
This updates ``inputs_embeds`` in place.
"""
## --
## really bad WA for placeholders - to fix
missmatch = is_multimodal.sum().item() - _flatten_embeddings(multimodal_embeddings).shape[0]

Check failure on line 365 in vllm/model_executor/models/utils.py

View workflow job for this annotation

GitHub Actions / ruff (3.12)

Ruff (E501) 

vllm/model_executor/models/utils.py:365:81: E501 Line too long (96 > 80)
if missmatch:
flat_is_multimodal = is_multimodal.flatten()
true_indices = torch.where(flat_is_multimodal == True)[0]

Check failure on line 368 in vllm/model_executor/models/utils.py

View workflow job for this annotation

GitHub Actions / ruff (3.12)

Ruff (E712) 

vllm/model_executor/models/utils.py:368:36: E712 Avoid equality comparisons to `True`; use `if flat_is_multimodal:` for truth checks
indices_to_remove = true_indices[-missmatch:] if missmatch > 0 else []
flat_is_multimodal[indices_to_remove] = False
is_multimodal = flat_is_multimodal.reshape(is_multimodal.shape)
## --
num_expected_tokens = is_multimodal.sum().item()
assert isinstance(num_expected_tokens, int)

Expand Down
Loading