Pytorch(二)：torch常用API应用及详解

目录

torch常用API

1.torch.max(x,dim=1)

1.1定义

1.2example

---

torch常用API

1.torch.max(x,dim=1)

1.1定义

• input输入的是一个tensor

• dim是max函数索引的维度0/1，0是每列的最大值，1是每行的最大值

• 返回的是两个值：一个是每一行最大值的tensor组，另一个是最大值所在的位置

max_col_value = torch.max(x,dim=0)[0]     # 每一列最大值
max_row_value = torch.max(x,dim=1)[0]     # 每一行最大值

1.2example

x:
 tensor([[0.5285, 0.1247, 0.8332, 0.5485],
        [0.7917, 0.6138, 0.5881, 0.3381],
        [0.4226, 0.6605, 0.8571, 0.0399],
        [0.1716, 0.0609, 0.9712, 0.4838]])
        
torch.max(x,1):
 (tensor([0.8332, 0.7917, 0.8571, 0.9712]), tensor([2, 0, 2, 2]))
 
torch.max(x,0):
 (tensor([0.7917, 0.6605, 0.9712, 0.5485]), tensor([1, 2, 3, 0]))
 
torch.max(x,1)[0]:
 tensor([0.8332, 0.7917, 0.8571, 0.9712])
 
torch.max(x,1)[1]:
 tensor([2, 0, 2, 2])
 
torch.max(x,1)[1].data:
 tensor([2, 0, 2, 2])
 
torch.max(x,1)[1].data.numpy():
 [2 0 2 2]
 
torch.max(x,1)[1].data.numpy().squeeze():
 [2 0 2 2]
 
torch.max(x,1)[0].data:
 tensor([0.8332, 0.7917, 0.8571, 0.9712])
 
torch.max(x,1)[0].data.numpy():
 [0.83318216 0.7917127  0.85708565 0.9711726 ]
 
torch.max(x,1)[0].data.numpy().squeeze():
 [0.83318216 0.7917127  0.85708565 0.9711726 ]