lucid.unsqueeze¶

lucid.unsqueeze(a: Tensor, /, axis: list[int] | tuple[int]) → Tensor¶

The unsqueeze function adds a dimension of size 1 at the specified axis, which can be useful for broadcasting in operations.

Function Signature¶

def unsqueeze(a: Tensor, axis: _ShapeLike) -> Tensor

Parameters¶

a (Tensor): The input tensor to unsqueeze.
axis (_ShapeLike): The axis along which to add the new dimension of size 1.

Returns¶

Tensor:
A tensor with an added dimension of size 1. If a requires gradients, the resulting tensor will also require gradients.

Forward Calculation¶

The unsqueeze operation adds a new dimension of size 1 at the specified axis.

\[\mathbf{out}_i = \text{unsqueeze}(\mathbf{a}_i, \text{axis})\]

Backward Gradient Calculation¶

For an unsqueezed tensor, the gradient is propagated through the new dimension without any changes to the values of the original tensor.

\[\frac{\partial \mathbf{out}}{\partial \mathbf{a}} = \mathbf{I}\]

Example¶

>>> import lucid
>>> a = Tensor([1.0, 2.0, 3.0], requires_grad=True)
>>> unsqueezed = lucid.unsqueeze(a, 0)  # or a.unsqueeze(0)
>>> print(unsqueezed)
Tensor([[1. 2. 3.]], grad=None)

Note

The unsqueeze operation is typically used to add a dimension to match broadcasting requirements in operations.