Repository navigation
fix(screenspot): specify normalized click coordinate units - #1761
Open
rudycelekli wants to merge 1 commit into
Open
rudycelekli wants to merge 1 commit into
rudycelekli wants to merge 1 commit into
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
ScreenSpot default prompts ask for x/y without stating the coordinate units, while the evaluator uses fractions of image dimensions. State that x is a fraction of image width and y of image height, both between 0 and 1, in the functional and composite user templates. This addresses the explicit prompt remedy proposed in #1706.
Scoring, pixel fallback, custom-model prompts and stored predictions are unchanged. No rescoring of saved 1000-grid results or measured model-accuracy gain is claimed. Related open #1719/#1722 change weighting and parsing, not these instructions.
Frozen baseline: four prompt regressions fail and one normalized-point control passes. Candidate and independent review: seven tests pass, including two image-helper controls. Fixtures execute actual builder/conversion/parser methods with real pandas rows and a local image, while isolating the heavy dataset import. New-test lint and whitespace pass; source diagnostics decrease from six to four with none introduced. ADR included in this reviewed revision. No full-package or model execution. AI-assisted contribution.