Skip to content

Fix the issue of blip model returning loss even when the label is not provided. #21811

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
sgugger merged 6 commits into from
Feb 28, 2023
Merged

Fix the issue of blip model returning loss even when the label is not provided. #21811

Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
f5b6370
Fix the issue of blip model returning loss even when the label is not…
raghavanone Feb 27, 2023
8b241d4
Fix ruff failure
raghavanone Feb 27, 2023
0129647
Incorporate PR feedbacks
raghavanone Feb 27, 2023
087d90b
Incorporate PR feedbacks
raghavanone Feb 27, 2023
001073b
Incorporate PR feedbacks
raghavanone Feb 27, 2023
ce7128e
Incorporate PR feedbacks
raghavanone Feb 27, 2023
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
8 changes: 1 addition & 7 deletions src/transformers/models/blip/modeling_blip.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -990,7 +990,7 @@ def forward(

>>> outputs = model(**inputs)
```"""
batch_size = pixel_values.shape[0]
pixel_values.shape[0]

@NielsRogge NielsRogge Feb 27, 2023

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this line can be removed as well

return_dict = return_dict if return_dict is not None else self.config.use_return_dict

vision_outputs = self.vision_model(
Expand All @@ -1002,12 +1002,6 @@ def forward(

image_embeds = vision_outputs[0]

if input_ids is None:
input_ids = torch.LongTensor([[self.decoder_input_ids] * batch_size]).to(image_embeds.device)

if labels is None:
labels = input_ids.masked_fill(input_ids == self.decoder_pad_token_id, -100)

outputs = self.text_decoder(
input_ids=input_ids,
attention_mask=attention_mask,
Expand Down