PyTorch .log1p()

In PyTorch, .log1p() computes the natural logarithm of one plus the input tensor element-wise.

This function improves numerical stability for small input values, avoiding issues that can occur when computing the logarithm directly.

Syntax

torch.log1p(input, *, out=None)

Parameters:

• input: The input tensor for which the natural logarithm of one plus the elements will be computed.
• out (Optional): A tensor to store the result. If provided, the computed values will be written to this tensor instead of creating a new one.

Return value:

Returns a new tensor where each element is the natural logarithm of one plus the corresponding element in the input tensor. The original tensor remains unchanged unless an out tensor is provided.

Example

In this example, .log1p() computes the natural logarithm of one plus each element in a tensor:

import torch

# Create a tensor
x = torch.tensor([0.0, 1.0, 2.0])

# Compute log1p
y = torch.log1p(x)

print(y)

Copy to clipboard
Copy to clipboard

The code above generates the output as follows:

tensor([0.0000, 0.6931, 1.0986])

Copy to clipboard
Copy to clipboard

In this example, the input tensor x contains the values [0.0, 1.0, 2.0]. The log1p() function computes the natural logarithm of one plus each element in the tensor, resulting in the output tensor [0.0000, 0.6931, 1.0986].