deep \\\\&cross network pytorch代码

深度交叉网络（Deep Cross Network, DCN）是一种用于显式学习交叉特征的模型。它由两部分组成：Cross Network和Deep Network。其中，Cross Network被用来高效地学习交叉特征，而Deep Network则用来学习深层次的非线性特征。

以下是一个使用PyTorch实现的深度交叉网络的代码示例：

import torch
import torch.nn as nn

class CrossNetwork(nn.Module):
    def __init__(self, layer_num, input_dim):
        super(CrossNetwork, self).__init__()
        self.layer_num = layer_num
        self.cross_weights = nn.ParameterList([nn.Parameter(torch.rand(input_dim, 1)) for _ in range(layer_num)])
        self.cross_bias = nn.ParameterList([nn.Parameter(torch.rand(input_dim, 1)) for _ in range(layer_num)])

    def forward(self, x):
        x_0 = torch.unsqueeze(x, dim=2)
        x = x_0.clone()
        x_T = x_0.clone().permute((0, 2, 1))

        for i in range(self.layer_num):
            x = torch.matmul(torch.bmm(x_0, x_T), self.cross_weights[i]) + self.cross_bias[i]
            x_T = x.clone().permute((0, 2, 1))

        x = torch.squeeze(x)
        return x

class DeepCrossNetwork(nn.Module):
    def __init__(self, input_dim, cross_layer_num, deep_layer_sizes):
        super(DeepCrossNetwork, self).__init__()
        self.cross_network = CrossNetwork(cross_layer_num, input_dim)
        self.deep_network = nn.Sequential(