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

Reduce img size in dummy_inputs for FLUX #1070

Merged
merged 3 commits into from
Dec 17, 2024
+3 −2
Merged

Reduce img size in dummy_inputs for FLUX #1070

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d99b690
Reduce img size in dummy_inputs for FLUX
mvafin Dec 12, 2024
0850f4c
change dims in function signature
IlyasMoutawwakil Dec 17, 2024
adfdc68
Update optimum/exporters/openvino/model_configs.py
IlyasMoutawwakil Dec 17, 2024
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
5 changes: 3 additions & 2 deletions optimum/exporters/openvino/model_configs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1804,8 +1804,9 @@ def __init__(
normalized_config: NormalizedVisionConfig,
batch_size: int = DEFAULT_DUMMY_SHAPES["batch_size"],
num_channels: int = DEFAULT_DUMMY_SHAPES["num_channels"],
width: int = DEFAULT_DUMMY_SHAPES["width"],
height: int = DEFAULT_DUMMY_SHAPES["height"],
width: int = DEFAULT_DUMMY_SHAPES["width"] // 4,
Copy link
Contributor Author

Choose a reason for hiding this comment

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

This unfortunately doesn't work as defaults are set in the optimum code and this will not change the behaviour

Copy link
Member

Choose a reason for hiding this comment

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

oh ! damn, then it's gonna be problematic as we also don't want to divide user specified input shapes

Copy link
Member

Choose a reason for hiding this comment

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

@echarlaix do you know where exactly we can change these values for a specific architecture like Flux ?
I thought dummy inputs generators will use the defaults in their signature by default unless a user passes input shapes.

height: int = DEFAULT_DUMMY_SHAPES["height"] // 4,
# Reduce img shape by 4 for FLUX to reduce memory usage on conversion
**kwargs,
):
super().__init__(task, normalized_config, batch_size, num_channels, width, height, **kwargs)
Expand Down