cnn IndexError：目标 2 超出范围

Question

执行代码后出现此错误，似乎代码的以下部分引发了此错误。我尝试了不同的方法，但没有什么能解决它。误差由损失函数给出。

for i, data in enumerate(train_loader, 0):


      #  import pdb;pdb.set_trace()
        inputs, labels = data
        print(type(inputs))
        for input in inputs:
            inputs = torch.Tensor(input)
        inputs, labels= Variable(inputs), Variable(labels)
        inputs=inputs.unsqueeze(1)
        optimizer.zero_grad()
        outputs = net(inputs)
        #import pdb;pdb.set_trace()
        loss_size = loss(outputs, labels)
        loss_size.backward()
        optimizer.step()

        running_loss += loss_size.data[0]
        total_train_loss += loss_size.data[0]

        if (i + 1) % (print_every + 1) == 0:
            print("Epoch {}, {:d}% \t train_loss: {:.2f} took: {:.2f}s".format(
                    epoch+1, int(100 * (i+1) / n_batches), running_loss / print_every, time.time() - start_time))
            running_loss = 0.0
            start_time = time.time()

--------------------------------------------------------------------------- IndexError                                Traceback (most recent call
last) <ipython-input-10-7d1b8710defa> in <module>
      1 CNN = Net()
----> 2 trainNet(CNN, learning_rate=0.001)
      3 #test()

<ipython-input-7-3208c0794681> in trainNet(net, learning_rate)
     23             outputs = net(inputs)
     24             #import pdb;pdb.set_trace()
---> 25             loss_size = loss(outputs, labels)
     26             loss_size.backward()
     27             optimizer.step()

~\Documents\Anaconda3\lib\site-packages\torch\nn\modules\module.py in
__call__(self, *input, **kwargs)
    530             result = self._slow_forward(*input, **kwargs)
    531         else:
--> 532             result = self.forward(*input, **kwargs)
    533         for hook in self._forward_hooks.values():
    534             hook_result = hook(self, input, result)

~\Documents\Anaconda3\lib\site-packages\torch\nn\modules\loss.py in
forward(self, input, target)
    914     def forward(self, input, target):
    915         return F.cross_entropy(input, target, weight=self.weight,
--> 916                                ignore_index=self.ignore_index, reduction=self.reduction)
    917 
    918 

~\Documents\Anaconda3\lib\site-packages\torch\nn\functional.py in
cross_entropy(input, target, weight, size_average, ignore_index,
reduce, reduction)    2019     if size_average is not None or reduce
is not None:    2020         reduction =
_Reduction.legacy_get_string(size_average, reduce)
-> 2021     return nll_loss(log_softmax(input, 1), target, weight, None, ignore_index, None, reduction)    2022     2023 

~\Documents\Anaconda3\lib\site-packages\torch\nn\functional.py in
nll_loss(input, target, weight, size_average, ignore_index, reduce,
reduction)    1836                          .format(input.size(0),
target.size(0)))    1837     if dim == 2:
-> 1838         ret = torch._C._nn.nll_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index)    1839     elif dim == 4:    1840         ret = torch._C._nn.nll_loss2d(input, target,
weight, _Reduction.get_enum(reduction), ignore_index)
IndexError: Target 2 is out of bounds.

IndexError: 目标 2 超出范围。

Answer 1

我遇到了同样的问题。问题是通过改变类的数量来解决的。

num_classes = 10（改为实际班级编号，而不是1）

Answer 2

您应该更改类数 = 3。
您可能将 1 和 2 作为类标签，因此您必须尝试将我们的模型网络类中的输出数量设置为 2，但它应该是 3，因为这是 pytorch 的工作方式。 2 类意味着您将 0 和 1 作为类标签。但是由于您将 1,2 作为类标签，因此您应该将其设为 3 类 (0,1,2) 分类问题。 假设这是您的网络课程：

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()

        self.layer_1 = nn.Linear(100, 10)
        self.layer_2 = nn.Linear(10, 2)

    def forward(self, x):
        x = self.layer_1(x)
        x = nn.relu(x)
        x = self.layer_2(x)
        x = nn.relu(x)

        return x

所以，你只需修改 layer_2 如下：
self.layer_2 = nn.Linear(10, 3)
这应该可以。

Answer 3

问题与您的类标签的值有关。您可以在这里找到一个很好的解释和解决方案： IndexError: Target is out of bounds