Repository navigation
feat: gate candidate fixtures at the supported game version floor - #42
Merged
Merged
Conversation
Fixtures from game builds below minimum_released_version in the compatibility config are no longer scored by the candidate comparison, so unsupported 5/6 month screenshots cannot fail a model that reads supported builds.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The candidate-vs-stable comparison scored every fixture in
datasets/fixtures/challenge, including 38 screenshots from 5/6 month builds that are belowminimum_released_version(v26.0811.1). A candidate trained on supported builds failed per-field no-regress on those.batch_eval.evaluatetakes an optionalminimum_version;compare_candidatepasses the floor fromconfigs/bastion_screenshot_compatibility.json. Fixture data and the legacyHOLDOUT_IDSflow are untouched; stable and candidate are filtered identically.Re-evaluating the 40aff1 Kaggle checkpoint under the floor: the four
versionregressions disappear; the only remaining regression isviewer_player21/22 vs 22/22 (recent_hud_17, 毁 read as 毀), a real model misread.