出现在一行代码中间位置的@这个@是矩阵相乘的意思， 就是比如 A的shape是[batch_size, sequence_length2, sequence_length1] B的shape是[batch_size, sequence_length1, hidden_dim] C = A @ B 则C的shape是[batch_size, s
出现在一行代码中间位置的@这个@是矩阵相乘的意思，
就是比如
A的shape是[batch_size, sequence_length2, sequence_length1]
B的shape是[batch_size, sequence_length1, hidden_dim]
C = A @ B
则C的shape是[batch_size, sequence_length2, hidden_dim]