pytorch函数torch.nn.functional.normalize()_torch.nn.functional.normalize dim=-1-优快云博客

本文链接：https://blog.youkuaiyun.com/qq_25603827/article/details/127146117

本文通过一个具体的代码示例展示了如何在PyTorch中使用不同维度进行张量归一化操作，并提供了归一化后的张量输出结果，有助于读者理解归一化在深度学习中的应用。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

import torch


a = torch.ones([1, 2, 3, 4])

a_0 = torch.nn.functional.normalize(a, dim=0)
a_1 = torch.nn.functional.normalize(a, dim=1)
a_2 = torch.nn.functional.normalize(a, dim=2)
a_3 = torch.nn.functional.normalize(a, dim=-1)
print(a)
print("***********")
print(a_0)
print("***********")
print(a_1)
print("***********")
print(a_2)
print("***********")
print(a_3)

输出结果：

tensor([[[[1., 1., 1., 1.],
          [1., 1., 1., 1.],
          [1., 1., 1., 1.]],

         [[1., 1., 1., 1.],
          [1., 1., 1., 1.],
          [1., 1., 1., 1.]]]])
***********
tensor([[[[1., 1., 1., 1.],
          [1., 1., 1., 1.],
          [1., 1., 1., 1.]],

         [[1., 1., 1., 1.],
          [1., 1., 1., 1.],
          [1., 1., 1., 1.]]]])
***********
tensor([[[[0.7071, 0.7071, 0.7071, 0.7071],
          [0.7071, 0.7071, 0.7071, 0.7071],
          [0.7071, 0.7071, 0.7071, 0.7071]],

         [[0.7071, 0.7071, 0.7071, 0.7071],
          [0.7071, 0.7071, 0.7071, 0.7071],
          [0.7071, 0.7071, 0.7071, 0.7071]]]])
***********
tensor([[[[0.5774, 0.5774, 0.5774, 0.5774],
          [0.5774, 0.5774, 0.5774, 0.5774],
          [0.5774, 0.5774, 0.5774, 0.5774]],

         [[0.5774, 0.5774, 0.5774, 0.5774],
          [0.5774, 0.5774, 0.5774, 0.5774],
          [0.5774, 0.5774, 0.5774, 0.5774]]]])
***********
tensor([[[[0.5000, 0.5000, 0.5000, 0.5000],
          [0.5000, 0.5000, 0.5000, 0.5000],
          [0.5000, 0.5000, 0.5000, 0.5000]],

         [[0.5000, 0.5000, 0.5000, 0.5000],
          [0.5000, 0.5000, 0.5000, 0.5000],
          [0.5000, 0.5000, 0.5000, 0.5000]]]])