normalize

Functions:

Name	Description
batchwise_min_max_normalize	Normalize a tensor by its min and max values within each batch element. This is equivalent to using
min_max_normalize	Normalize the given tensor by subtracting the minimum value and dividing by the range of its minimum value to its maximum value.

batchwise_min_max_normalize

batchwise_min_max_normalize(tensor: Tensor) -> <class 'torch.Tensor'>

Normalize a tensor by its min and max values within each batch element. This is equivalent to using min_max_normalize in a loop over the batch dimension.

Examples:

>>> tensor = torch.tensor([[0, 1, 2], [3, 4, 5]], dtype=torch.float32)
>>> batchwise_min_max_normalize(tensor)
tensor([[0.0000, 0.5000, 1.0000],
        [0.0000, 0.5000, 1.0000]])

Parameters:

Name	Type	Description	Default
tensor	Tensor	The tensor to normalize.	required

Returns:

Type	Description
<class 'torch.Tensor'>	The tensor, normalized batchwise.

Note

If the min and max values are the same along the batch dimension, nan values will be returned for that batch element.

min_max_normalize

min_max_normalize(tensor: Tensor) -> <class 'torch.Tensor'>

Normalize the given tensor by subtracting the minimum value and dividing by the range of its minimum value to its maximum value. This is equivalent to torchvision.utils.make_grid(normalize=True). If the range is zero, the tensor is clamped to [0, 1].

Examples:

>>> tensor = torch.tensor([[0, 1, 2], [3, 4, 5]], dtype=torch.float32)
>>> min_max_normalize(tensor)
tensor([[0.0000, 0.2000, 0.4000],
        [0.6000, 0.8000, 1.0000]])

Parameters:

Name	Type	Description	Default
tensor	Tensor	The tensor to normalize.	required

Returns:

Type	Description
<class 'torch.Tensor'>	The normalized tensor.