PyTorch torch.clamp() method clamps all the input elements into the range [ min, max ] and return a resulting tensor.

Syntax: torch.clamp(inp, min, max, out=None) Arguments

• inp: This is input tensor.
• min: This is a number and specifies the lower-bound of the range to which input to be clamped.
• max: This is a number and specifies the upper-bound of the range to which input to be clamped.
• out: The output tensor.

Return: It returns a Tensor.

Let's see this concept with the help of few examples: Example 1:

 

# Importing the PyTorch library 

import torch 

# A constant tensor of size n

a = torch.randn(6)

print(a)

​

# Applying the clamp function and 

# storing the result in 'out'

out = torch.clamp(a, min = 0.5, max = 0.9)

print(out)

Output:

 -0.9214
-0.1268
 1.1570
-0.2753
-0.0746
 0.7957
[torch.FloatTensor of size 6]
 0.5000
 0.5000
 0.9000
 0.5000
 0.5000
 0.7957
[torch.FloatTensor of size 6]

Example 2:

 

# Importing the PyTorch library 

import torch 

# A constant tensor of size n

a = torch.FloatTensor([1, 4, 6, 8, 10, 14])

print(a)

​

# Applying the clamp function and 

# storing the result in 'out'

out = torch.clamp(a, min = 5, max = 10)

print(out)

Output:

 1
  4
  6
  8
 10
 14
[torch.FloatTensor of size 6]
  5
  5
  6
  8
 10
 10
[torch.FloatTensor of size 6]?