Repository navigation
Audit frozen BBBC039 paired quality mechanisms without rerunning science - #798
Merged
Merged
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Completed coordinator-only frozen evidence review following #796; source pin 29898d2. Changes ONE receipt, no production/skill/renderer edits.
Both finals actually use MCE + SHAPE. Installed declarations confirm intensity-marker smoothing does not affect SHAPE; support smoothing, pre-fill timing, manual suppression and equivalent-area admission are interacting changed contracts. Installed morphology also differs, so no isolated parameter-causality claim.
Personally opened and hash-pinned 17 original native PNGs. Shared 20630_A02_1 shows changed local partitions with coherent isolated controls; within-run triples match, cross-run zoom/windows are not identical. The unequal A21 pair is admitted but lacks an obvious independent weaker seed; failed repairs preserved. A21 improves in whole-field annotation F1 and is not mislabeled a pooled regression. Largest scored decline 20589_P23_7 has no saved native matched triple.
Existing stage-specific learning/measurement/segmentation guidance already covers this decision, and relevant guides were retrieved before FIRST: no duplicate paragraph proposed. Exact cause of the pooled 314 additional FN/+11 FP remains unresolved. Actual #796 same200 JSON micro-F1 is 90.62244695% -> 89.83683813%; no rescoring.
Qualification: retained reports, actual installed source relationships, original route/camera/state receipts and 17 original PNGs read; SHA pins recorded; git diff --check passes. No tests/scorer/runtime/scientific execution, references opened as pixels, author feedback, new environment or worktree. Foreign gitlinks/untracked evidence preserved. #797 manuscript and Planck #748/#522 receiving remain separate. CI deferred.