manifold-power-iteration-e2d33f6a·1 events·first seen Aliases: Manifold Power Iteration
A new arXiv preprint proposes Manifold Power Iteration (MPI), a principled redesign of Mixture-of-Experts router matrices that aligns each router row with the principal singular direction of its associated expert. The method uses a 'Power-then-Retract' paradigm to enforce norm constraints while driving convergence toward these singular directions. Empirical validation spans MoE pretraining at scales from 1B to 11B parameters, showing improved model effectiveness.