为何运行PyTorch代码时出现IndexError: Target 1 is out of bounds错误？

问题重现 #

执行以下代码时，第一段正常运行，第二段设置target_tensor = [1]会触发索引越界错误：

import torch
import torch.nn as nn

# 第一段：正常执行
input_tensor = torch.tensor([[2.0]])
target_tensor = torch.tensor([0], dtype=torch.long)
loss_function = nn.CrossEntropyLoss()
loss = loss_function(input_tensor, target_tensor)
print(loss)

# 第二段：触发错误
input_tensor = torch.tensor([[2.0]])
target_tensor = torch.tensor([1], dtype=torch.long)
loss_function = nn.CrossEntropyLoss()
loss = loss_function(input_tensor, target_tensor)
print(loss)

错误栈信息：

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
Cell In[214], line 14
     12 target_tensor = torch.tensor([1], dtype=torch.long)
     13 loss_function = nn.CrossEntropyLoss()
---> 14 loss = loss_function(input_tensor, target_tensor)
     15 print(loss)

File ~/jupyter-env-3.10/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/jupyter-env-3.10/lib/python3.10/site-packages/torch/nn/modules/loss.py:1163, in CrossEntropyLoss.forward(self, input, target)
   1162 def forward(self, input: Tensor, target: Tensor) -> Tensor:
-> 1163     return F.cross_entropy(input, target, weight=self.weight,
   1164                            ignore_index=self.ignore_index, reduction=self.reduction,
   1165                            label_smoothing=self.label_smoothing)

File ~/jupyter-env-3.10/lib/python3.10/site-packages/torch/nn/functional.py:2996, in cross_entropy(input, target, weight, size_average, ignore_index, reduce, reduction, label_smoothing)
   2994 if size_average is not None or reduce is not None:
   2995     reduction = _Reduction.legacy_get_string(size_average, reduce)
-> 2996 return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)

IndexError: Target 1 is out of bounds.

核心原因 #

PyTorch的CrossEntropyLoss对输入有明确要求：

• 输入张量input的形状必须是[batch_size, num_classes]，其中num_classes是分类任务的类别总数
• 目标张量target中的每个元素必须是0到num_classes-1之间的整数，代表样本所属的类别索引

你的代码中，input_tensor = torch.tensor([[2.0]])的形状是[1,1]，意味着num_classes=1，此时合法的目标索引只能是0。当你把target设为1时，超出了0~0的范围，直接触发索引越界错误。

解决方法 #

根据你的实际任务需求选择对应方案：

• 如果是二分类任务：将输入张量调整为包含两个类别的预测值，比如input_tensor = torch.tensor([[2.0, 1.0]])，此时num_classes=2，target可以设为0或1。
• 如果是单分类/二分类的另一种实现：若任务本质是判断样本是否属于某一类，可改用BCELoss（二分类交叉熵损失），注意需要在模型输出后添加sigmoid激活，同时target的取值应为0或1的浮点型张量。
• 确认任务类别数：若确实只有1个类别，那target只能设为0，无需调整输入形状。

内容的提问来源于stack exchange，提问作者Sazzad Hissain Khan