Fix image resize minimum-size handling#358
Open
Travor278 wants to merge 1 commit into
Open
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
Fixes the image minimum-size resize path used during GRPO generation scoring. Normal-size images are now returned unchanged, while images with either side below 28 pixels are resized once with aspect ratio preserved.
The trainer behavior for valid images is unchanged except that it no longer relies on an
UnboundLocalError/broad exception path for normal images. A small image utility and focused tests cover the resize cases.Related Issue
Resolves #355.
Motivation and Context
The previous code defined
new_wandnew_honly inside the small-image branch but calledimg.resize((new_w, new_h), ...)outside that branch. For normal images, this triggered exception handling as part of normal control flow. Moving the behavior into a tested helper makes the intended resize contract explicit and avoids that overhead.How Has This Been / Can This Be Tested?
Tested locally on Windows with Anaconda
py311/ Python 3.11.15.The focused image tests passed:
5 passed.Checklist