Repository navigation
fix: use platform-measured query encoder baseline - #53
Merged
Merged
Conversation
Keep reference implementation and input fingerprints pinned while accepting the current verifier's measured quality baseline. Validate RewardKit audit details so internal judge failures invalidate the measurement, and add offline regression coverage and reevaluation guidance.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 query-encoder-alignment verifier already measures its quality baseline locally, but previously rejected measurements that differed from the frozen A100 calibration. Use the current run’s measured baseline while retaining implementation and input fingerprint checks.
This also validates RewardKit result details so internal judge timeouts are recorded as invalid measurements, and documents verifier-only reevaluation using the original submission and trajectory.
The scoring formula, relative runtime gate, datasets and task constraints remain unchanged.
Validation: