生成堆叠降噪自编码器的pytorch代码

以下是生成堆叠降噪自编码器的PyTorch代码：

import torch
import torch.nn as nn

class DenoisingAutoencoder(nn.Module):
    def __init__(self, input_dim, hidden_dims):
        super(DenoisingAutoencoder, self).__init__()
        self.input_dim = input_dim
        self.hidden_dims = hidden_dims
        self.num_layers = len(hidden_dims)
        
        # Encoder layers
        self.encoder_layers = nn.ModuleList()
        for i in range(self.num_layers):
            if i == 0:
                self.encoder_layers.append(nn.Linear(input_dim, hidden_dims[i]))
            else:
                self.encoder_layers.append(nn.Linear(hidden_dims[i-1], hidden_dims[i]))
            self.encoder_layers.append(nn.ReLU(inplace=True))
            
        # Decoder layers
        self.decoder_layers = nn.ModuleList()
        for i in range(self.num_layers-1, -1, -1):
            if i == self.num_layers-1:
                self.decoder_layers.append(nn.Linear(hidden_dims[i], input_dim))
            else:
                self.decoder_layers.append(nn.Linear(hidden_dims[i+1], hidden_dims[i]))
            self.decoder_layers.append(nn.ReLU(inplace=True))
        
        self.loss = nn.MSELoss()
    
    def forward(self, x):
        # Add noise to input
        noise = torch.randn(x.shape)
        noisy_x = x + noise
        
        # Encoder
        for layer in self.encoder_layers:
            x = layer(x)
        
        # Decoder
        for layer in self.decoder_layers:
            x = layer(x)
        
        return x, noisy_x
    
    def compute_loss(self, x, pred_x):
        loss = self.loss(x, pred_x)
        return loss

该代码定义了一个堆叠降噪自编码器（stacked denoising autoencoder），它具有多个隐藏层。输入维度和隐藏层维度作为参数传递给模型。模型包含一个编码器和一个解码器。编码器将输入转换为隐藏表示，而解码器将隐藏表示转换回输入。在训练期间，模型使用带有随机噪声的输入进行训练，以帮助模型学习具有鲁棒性的隐藏表示。该模型使用均方误差损失函数来计算重构损失。