arXiv stat.ML
· Papers
FlowAdam: Implicit Regularization via Geometry-Aware Soft Momentum Injection
arXiv:2604.06652v1 Announce Type: cross Abstract: Adaptive moment methods such as Adam use a diagonal, coordinate-wise preconditioner based on exponential moving averages of squared gradients. This diagonal scaling is coordinate-system dependent and can struggle with dense or rotated parameter couplings, including thos