AWhat changed
Used maximal-update parameterization to tune optimization/initialization settings on smaller neural networks and transfer them to larger models without direct full-scale tuning.
Scientific scope
Used maximal-update parameterization to tune optimization/initialization settings on smaller neural networks and transfer them to larger models without direct full-scale tuning. Inherits muP and tensor-program theory. Does not claim invention of muP or transfer of every hyperparameter: original table excludes regularization such as dropout/weight decay, while depth/batch/sequence/time transfer has empirical caveats. Small-model tuning still costs compute.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Inherits muP and tensor-program theory. Does not claim invention of muP or transfer of every hyperparameter: original table excludes regularization such as dropout/weight decay, while depth/batch/sequence/time transfer has empirical caveats. Small-model tuning still costs compute.
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