- tf.multiply : 按元素相乘
等于torch.mul()
# [64*6*30, 2] * [64*6*30,1]
this_platf_target_outputs = tf.multiply(neighbor_part_inputs, mask[target_node, platf, ...])
Returns x * y element-wise
- tf.matmul: 矩阵相乘
this_platf_target_outputs = tf.matmul(this_platf_target_outputs, self.wc[platf]) + self.bc[platf] # shape=(seq_len*num_nodes, units)