关于pytorch交叉熵报错:Dimension out of range (expected to be in range of [-1, 0], but got 1)
有博文说是因为数据类型的原因,排查以后发现,target数据类型是需要为float的,不然会继续报错:RuntimeError: Expected floating point type for target with class probabilities, got Long
问题实际上是出在output和target的维度上,需要输入一个二维即以上的数据维度,所以当数据为一维向量的时候,加一个维度就好
以下是测试代码:
import torch.nn as nn
criterion = nn.CrossEntropyLoss()
output = torch.rand(3)
target = torch.rand(3)
loss = criterion(output, target)
IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
output = torch.rand(1,3)
target = torch.rand(1,3)
loss = criterion(output, target)