Fixed EfficientNet interpolation method to match the original implementation #42860
+586
−8
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ntation in the code Original method --> Bicubic\ Source: tpu/models/official/efficientnet/preprocessing.py Function def _resize_image() Line 88 - Changed default resample from Nearest to BICUBIC - Original Tensorflow implementation uses resize_bicubic by default in original code - Fixes Issue huggingface#28180
- Reorder imports in metaclip_2 conversion script to satisfy ruff formatter - Remove -> None annotation from EfficientNetImageProcessor.__init__ to match docstring checker expectations
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: efficientnet, metaclip_2 |
yonigozlan
reviewed
Dec 18, 2025
Member
yonigozlan
left a comment
yonigozlan
left a comment
There was a problem hiding this comment.
Happy to merge this as it is indeed the wrong default, however you also have to change it for the fast image processor!
| self.include_top = include_top | ||
|
|
||
| # Copied from transformers.models.vit.image_processing_vit.ViTImageProcessor.resize with PILImageResampling.BILINEAR->PILImageResampling.NEAREST | ||
| # Method for resize bicubic interpolation |
Member
There was a problem hiding this comment.
Suggested change
| # Method for resize bicubic interpolation |
|
|
||
| # Import MetaCLIP modules | ||
| from src.mini_clip.factory import create_model_and_transforms | ||
|
|
| @@ -0,0 +1,579 @@ | |||
| # Image Processor Interpolation Method Verification Guide | |||
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Fixes the default interpolation method for EfficientNet image processor from
NEARESTtoBICUBICto match the original TensorFlow implementation.Motivation
As pointed out in #28180 to check the models in the list, the EfficientNet image processor was using
NEARESTinterpolation , but the original TensorFlow implementation usesBICUBICby default.The
_resize_image()function in the original code explicitly usestf.image.resize_bicubicas the default resize method.Changes
resampleparameter fromPIL.Image.NEARESTtoPILImageResampling.BICUBICFixes #28180 (partial - for EfficientNet model)
Before submitting
cc @yonigozlan @molbap (vision models reviewers)