The preconditioning exponent p in Adam-like optimizers systematically shifts the relative update ratio between weight and bias parameters, changing which samples the model fits first.
Advances in Neural Information Processing Systems37, 23988–24021 (2024)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Information Allocation Dynamics in Neural Network Optimization
The preconditioning exponent p in Adam-like optimizers systematically shifts the relative update ratio between weight and bias parameters, changing which samples the model fits first.