【无标题】

Lyhlcs

已于 2023-03-12 18:44:01 修改

阅读量338

点赞数

文章标签：编辑器

于 2023-02-28 22:05:17 首次发布

本文链接：https://blog.csdn.net/Lyhlcs/article/details/129271092

版权

在运行PyTorch LSTM-CNN模型时遇到输入维度不匹配和嵌入层索引类型错误的问题。错误信息表明输入应当是3维，但实际上为4维，并且在使用embedding层时要求输入为Long类型而非Float。需要调整输入数据的形状并确保嵌入层的输入正确。

摘要由CSDN通过智能技术生成

最新问题
Traceback (most recent call last):
File “D:/pythonProject1/lstmcnn/lstmcnn.py”, line 178, in
output_ = model(x_input)
File “D:\Anaconda\lib\site-packages\torch\nn\modules\module.py”, line 1051, in _call_impl
return forward_call(*input, **kwargs)
File “D:/pythonProject1/lstmcnn/lstmcnn.py”, line 105, in forward
x = self.conv1(x) # 32, 50, 178
File “D:\Anaconda\lib\site-packages\torch\nn\modules\module.py”, line 1051, in _call_impl
return forward_call(*input, **kwargs)
File “D:\Anaconda\lib\site-packages\torch\nn\modules\conv.py”, line 298, in forward
return self._conv_forward(input, self.weight, self.bias)
File “D:\Anaconda\lib\site-packages\torch\nn\modules\conv.py”, line 294, in _conv_forward
return F.conv1d(input, weight, bias, self.stride,
RuntimeError: Expected 3-dimensional input for 3-dimensional weight [50, 20, 3], but got 4-dimensional input of size [180, 100, 32, 20] instead
输入的维度要怎么改呢

# 定义网络结构
class LSTM_CNN(nn.Module):
    def __init__(self, vocab_size, hidden_dim, num_layers, embedding_dim, output_dim):
        super(LSTM_CNN, self).__init__()
        self.hidden_dim = hidden_dim  # 隐层大小
        self.num_layers = num_layers  # LSTM层数
        # 嵌入层，会对所有词形成一个连续型嵌入向量，该向量的维度为embedding_dim
        # 然后利用这个向量来表示该字，而不是用索引继续表示
        self.embeddings = nn.Embedding(vocab_size + 1, embedding_dim)
        # 定义LSTM层，第一个参数为每个时间步的特征大小，这里就是每个字的维度
        # 第二个参数为隐层大小
        # 第三个参数为lstm的层数
        self.lstm = nn.LSTM(embedding_dim, hidden_dim, num_layers)
        self.conv1 = nn.Conv1d(embedding_dim, 50, 3)
        self.maxpool1 = nn.AdaptiveAvgPool1d(output_size=100)
        self.conv2 = nn.Conv1d(50, 100, 3)
        self.maxpool2 = nn.AdaptiveAvgPool1d(output_size=50)
        self.fc = nn.Linear(50 * 100, output_dim)

    def forward(self, x):
        # 1.首先形成嵌入向量
        embeds = self.embeddings(x)
        # 2.将嵌入向量导入到lstm层
        output, (h_n, c_n) = self.lstm(embeds)
        # timestep, batch_size, hidden_dim = output.shape
        # output = output.reshape(-1, hidden_dim)
        x = self.embeddings(output)  # 32,180, 16 批次，序列长度，词嵌入长度
        x = x.transpose(1, 2)  # 32, 16, 180 批次，词嵌入长度，序列长度

        x = self.conv1(x)  # 32, 50, 178
        x = self.maxpool1(x)  # 32, 50, 100
        x = self.conv2(x)  # 32, 100, 1