Python PyTorch clamp() Method

To clip each component in a contribution to the reach [min, max], call torch.clamp(). Three boundaries are required: the information tensor, the base, and the most extreme qualities. The qualities not exactly the min are supplanted by the min, and the qualities bigger than the maximum are supplanted by the maximum.

There is no lower bound on the off chance that min isn't given. There won't be an upper bound, assuming Max is forgotten about. Expecting that min and max are set to - 0.5 and 0.4, individually, values beneath - 0.5 or more 0.4 are subbed. There is no adjustment in the middle between these qualities. It just acknowledges inputs with genuine qualities.

Syntax

Parameters

Parameters

• input - The input tensor.
• min - Lower bound; it is a number or tensor.
• max - upper bound; it is a number or tensor.

After all input items are pinched into the range [min, max], a new tensor is returned.

Steps

• Bring in the necessary library. The necessary Python module for each of the next examples is a torch. Verify whether you have installed it already.

• Make an input tensor, then print it.

• Clamp the elements of the input tensor. Here, we use min=-0.5, max=0.5.

• Once the tensor has been clamped, print it.

Example 1:

The components of a 1D input tensor are clamped in the Python program that follows. When either min or max is None, take note of how the clamp() function operates.

Output:

 input tensor: tensor([ 0.7300, 0.3500, -0.3900, -1.5300]) clamp the tensor: into range [-0.5, 0.5]: tensor([ 0.5000, 0.3500, -0.3900, -0.5000]) if min is None: tensor([ 0.5000, 0.3500, -0.3900, -1.5300]) if max is None: tensor([0.7300, 0.5000, 0.5000, 0.5000]) if min is greater than max: tensor([0.5000, 0.5000, 0.5000, 0.5000]) 

Explanation:

This Python code illustrates how to limit a tensor's values inside certain constraints using PyTorch's torch.clamp() method. After generating a 1D tensor a, torch.clamp() is used to clamp the values in the interval [-0.5, 0.5]. Several examples are demonstrated, including managing situations where min is greater than max and clamping with both min and max bounds, max bounds alone, and min bounds only. The corresponding outputs for each case are shown, demonstrating how the tensor values are changed in response to the given restrictions.

Example 2

The components of a 2D input tensor are clamped in the Python program that follows. When either min or max is None, take note of how the clamp() function operates.

Output:

 input tensor: tensor([[-0.7272, -1.4916, -0.0681, -0.6933], [-0.3091, -1.3365, 1.6109, -0.8578], [ 0.7421, 0.6433, 1.6438, 2.3845]]) clamp the tensor: into range [-0.6, 0.4]: tensor([[-0.6000, -0.6000, -0.0681, -0.6000], [-0.3091, -0.6000, 0.4000, -0.6000], [ 0.4000, 0.4000, 0.4000, 0.4000]]) if min is None (max=0.4): tensor([[-0.7272, -1.4916, -0.0681, -0.6933], [-0.3091, -1.3365, 0.4000, -0.8578], [ 0.4000, 0.4000, 0.4000, 0.4000]]) if max is None (min=-0.6): tensor([[-0.6000, -0.6000, -0.0681, -0.6000], [-0.3091, -0.6000, 1.6109, -0.6000], [ 0.7421, 0.6433, 1.6438, 2.3845]]) if min is greater than max (min=0.6, max=0.4): tensor([[0.4000, 0.4000, 0.4000, 0.4000], [0.4000, 0.4000, 0.4000, 0.4000], [0.4000, 0.4000, 0.4000, 0.4000]]) 

Explanation:

This Python script shows how to constrain tensor data inside certain bounds using PyTorch's torch.clamp() function. In order to demonstrate scenarios with both minimum and maximum limits, only maximum bound, only minimum bound, and handling circumstances where the minimum bound surpasses the maximum, it constructs a 2D tensor and uses a torch.clamp() to limit values between -0.6 and 0.4. The result shows the appropriate adjustment of the tensor values.