Conversation
zzylol
changed the title
zzylol
marked this pull request as draft
September 28, 2026 02:28
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Deferred follow-up. Not required for the current #728 → #742 → #775 synthetic-selection and data-plane execution milestone. Online ERP feedback and runtime replanning are explicitly out of scope for that milestone.
Make accuracy-evidence admission independently reviewable from candidate pricing.
Before this PR: evidence applicability was mainly covered inside ERP implementation tests.
After this PR: public-adapter integration tests reject mismatched distribution, implementation, metric, parameters, insufficient trials and excessive error even for cheap candidates; distinguish empirical admission from theoretical/exact fallback; and retain admitted record identity.
Validation: three
accuracy_evidencetests passed. Records are explicitly synthetic contract fixtures, not calibration results. Existing readout-specific, formal-guarantee and freshness checks remain in place.Base: workload-statistics tests. Next: resource-measurement tests.