1 paper · 1 filter
Xianliang Li, Zihan Zhang, Weiyang Liu +1
Muon has recently demonstrated strong empirical performance in large language model training, but the theoretical role of momentum in Muon remains unclear. Existing analyses of Muo…