torch.nn.CrossEntropyLoss

原创于 2019-09-17 17:07:28 发布 · 1.3k 阅读

CC 4.0 BY-SA版权

文章标签：

torch.nn.CrossEntropyLoss

nn.logSoftmax()对input进行softmax操作
nn.NLLLoss()将softmax后的inout与target比较
nn.CrossEntropyLoss()是nn.logSoftmax()和nn.NLLLoss()的整合

import torch
import torch.nn as nn

input=torch.Tensor([[-0.1234, -0.2345,-0.3456]])
target = torch.tensor([0])

entroy=nn.CrossEntropyLoss()
output = entroy(input, target)
print(output)

tensor(0.9916)

m = nn.LogSoftmax()
loss = nn.NLLLoss()
input=m(input)
print(input)
output = loss(input, target)
print('output:',output)

tensor([[-0.9916, -1.1027, -1.2138]])
output: tensor(0.9916)