Why Deep Transformers are Difficult to Converge? From Computation Order to Lipschitz Restricted Parameter Initialization

@misc{pub12020,
    author = {

van Genabith, Josef
 and 

Xu, Hongfei
 and 

Liz, Qiuhui
 and 

Zhang, Jingyi
},
    title = {Why Deep Transformers are Difficult to Converge? From Computation Order to Lipschitz Restricted Parameter Initialization},
    publisher = {keine Angabe}
}