MHA-MLP算法流程图
2019-04-28 12:12:39 1 举报multi-head attention multi-layer perceptron
作者其他创作
大纲/内容
V
Linear
Concat
h
Max Pooling
Normalized NumericFeatures
Scaled Dot-Product Attention
Embedding
Output Y
Category Features
Dropout
. . .
Batch Normalization
Input X
K
Average Pooling
Q
MHA Layer
Repeat Vector
收藏
0 条评论
下一页
