pytorch CrossEntropyLoss用法

数据类型只支持long类型

只支持两个维度：

import torch 
import torch.nn as nn

loss = nn.CrossEntropyLoss()

# input, NxC=2x3
input = torch.randn(2, 3, requires_grad=True)

# target, N
target = torch.empty(2, dtype=torch.long).random_(3)
output = loss(input, target)
output.backward()

原文链接：https://blog.csdn.net/jacke121/article/details/83042423

多维度会报错：

1only batches of spatial targets supported (non-empty 3D tensors) but got targets of size: : [1, 3,

例如：

criterion = nn.CrossEntropyLoss()
 
loss = criterion(logit, target.long())
其中，logit: torch.Size([4, 9, 256, 256]);  target: [4,1, 256, 256]

就会出现1only batches of spatial targets supported (non-empty 3D tensors) but got targets of size的错误

错误原因：
Pytorch的CrossEntropy官方文档截图如下：

显然，其中N=4, C=31