6.1.4.1节中公式(6.1.3)有误
Author: LeatsyCreated May 17, 2026Updated May 17, 2026
文中所给公式是将attention score A(X)和V(X)进行element-wise乘法，论文'Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations'原文第三节 中给出的是一般性的矩阵乘法。况且如果此处为element-wise乘法，那其实attention机制在此处失去了本来的作用。
Source: datawhalechina/fun-rec