CNN Pytorch only batches of spatial targets supported 错误

Question

所以我构建了以下模型，我将使用它对 MNIST 时尚数据进行分类。

class CNN(nn.Module):
    def __init__(self, **kwargs):
      super().__init__()

  
      self.conv1    = nn.Conv2d(784, 64, 2, 1, padding=5)
      self.maxpool  = nn.MaxPool2d(kernel_size=2, stride=2)
      self.conv2    = nn.Conv2d(64, 128, 2, 2, padding = 0)
      self.conv2_bn = nn.BatchNorm2d(128)
      self.relu     = nn.ReLU()
      self.dense    = nn.Linear(1, 128)
      self.softmax  = nn.Softmax()

    def forward(self, x): # you can add any additional parameters you want
      
        x = self.conv1(x)
        x = F.max_pool2d(F.relu(x), kernel_size=2)
        x = self.conv2(x)
        x = self.conv2_bn(x)
        x = F.max_pool2d(F.relu(x), kernel_size=2)
        print(x.shape)
        x = self.dense(x)
        x = F.relu(x)

        return F.log_softmax(x)

这就是我运行我的代码：

for epoch in range(max_epoch):

    print('EPOCH='+str(epoch))

    correct = 0
    total = 0   
    
    running_loss = 0

    for data, label in tzip(TRAX, TRAY):

        #train = data.view(64,1,2,2)

        DAAA = data.view(1,784,1,1)
        #zeroing the parameter
        optimizer.zero_grad()
        label = torch.tensor([label]).type(torch.LongTensor)
        #forwards prop
        outputs = model2(DAAA)
        loss = criterion(outputs, label)
        loss.backward()
        optimizer.step()
        running_loss += loss.item()

        '========================================'
        _, predicted = torch.max(outputs.data, 1)
        total += label.size(0)
        correct += (predicted == label).sum().item()

        '========================================'
        
    print('\n')
    print('Accuracy of the network on the 10000 test images: %d %%' % (
    100 * correct / total))
    print('\n')
    print(str(epoch)+'loss= '+str(running_loss)) 

    lossjournal.append(running_loss)
    accjournal.append(100 * correct / total)

print('Finished Training')

<ipython-input-378-27ce013b2c10> in <module>
     55         #forwards prop
     56         outputs = model2(DAAA)
---> 57         loss = criterion(outputs, label)
     58         loss.backward()
     59         optimizer.step()

/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
    887             result = self._slow_forward(*input, **kwargs)
    888         else:
--> 889             result = self.forward(*input, **kwargs)
    890         for hook in itertools.chain(
    891                 _global_forward_hooks.values(),

/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch/nn/modules/loss.py in forward(self, input, target)
   1045     def forward(self, input: Tensor, target: Tensor) -> Tensor:
   1046         assert self.weight is None or isinstance(self.weight, Tensor)
-> 1047         return F.cross_entropy(input, target, weight=self.weight,
   1048                                ignore_index=self.ignore_index, reduction=self.reduction)
   1049 

/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch/nn/functional.py in cross_entropy(input, target, weight, size_average, ignore_index, reduce, reduction)
   2691     if size_average is not None or reduce is not None:
   2692         reduction = _Reduction.legacy_get_string(size_average, reduce)
-> 2693     return nll_loss(log_softmax(input, 1), target, weight, None, ignore_index, None, reduction)
   2694 
   2695 

/Library/Frameworks/Python.framework/Versions/3.9/lib/python3.9/site-packages/torch/nn/functional.py in nll_loss(input, target, weight, size_average, ignore_index, reduce, reduction)
   2388         ret = torch._C._nn.nll_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
   2389     elif dim == 4:
-> 2390         ret = torch._C._nn.nll_loss2d(input, target, weight, _Reduction.get_enum(reduction), ignore_index)
   2391     else:
   2392         # dim == 3 or dim > 4

RuntimeError: only batches of spatial targets supported (3D tensors) but got targets of dimension: 1

当我运行我的模型出现此错误但我不知道从现在开始该怎么做？我应该为这个模型做哪些调整？我知道问题出在标准上，但这是因为模型的输出形状为 [1, 128, 1, 128]?

Answer 1

MNIST 有 10 个类，因此您的输出大小应为 [batch_size, 10]。将最后一个线性层更改为 self.dense = nn.Linear(128,10)。然后，由于您的标签大小为 [batch_size,1]，因此您应该使用 torch.nn.CrossEntropyLoss 作为标准。

此外，您不需要在训练期间包括最后 softmax 层，因为上述损失函数在计算期间执行 softmax 操作。您可以改用 softmax 或 argmax 仅用于推理。

CNN Pytorch only batches of spatial targets supported 错误

CNN Pytorch only batches of spatial targets supported error

python

machine-learning

python-3.x

conv-neural-network

pytorch