AWhat changed
Accumulates gradient agreement to approximate a training point’s effect on test loss along optimization, with practical checkpoint and selected-layer variants for scalable inspection.
Scientific scope
Accumulates gradient agreement to approximate a training point’s effect on test loss along optimization, with practical checkpoint and selected-layer variants for scalable inspection. First-order gradient approximations and checkpoint heuristics estimate training-example influence along an optimization trajectory. They are not exact deletion-and-retraining effects; extending the method to another optimizer requires its actual update rule.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. First-order gradient approximations and checkpoint heuristics estimate training-example influence along an optimization trajectory. They are not exact deletion-and-retraining effects; extending the method to another optimizer requires its actual update rule.
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.
- Estimating Training Data Influence by Tracking Gradient Descent
- Estimating Training Data Influence by Tracing Gradient Descent
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