AWhat changed
Combined negatives across GPU batches, filtered likely false negatives with a cross-encoder and used high-confidence pseudo-labeled question/passage pairs in a staged dual-encoder training pipeline.
Scientific scope
Combined negatives across GPU batches, filtered likely false negatives with a cross-encoder and used high-confidence pseudo-labeled question/passage pairs in a staged dual-encoder training pipeline. Teacher scores are imperfect relevance estimates, not certified clean labels. Cross-device communication, teacher inference and staged retraining have costs. Snapshot leaderboard claims do not establish present ranking or all-domain superiority.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Teacher scores are imperfect relevance estimates, not certified clean labels. Cross-device communication, teacher inference and staged retraining have costs. Snapshot leaderboard claims do not establish present ranking or all-domain superiority.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction