openalexFigshare2026-07-25
MAS_ Multi-stage Adaptive Fusion Optimizer Combining Muon, AdamW and SGD
Adaptive optimizers constitute core components for training deep learning models.However, mainstream optimizers suffer from irreconcilable inherent flaws. AdamWtends to induce over-preconditioning due to long-term accumulation of second-ordermoments, trapping models in sharp loca…