nn.functional.max_pool3d¶

lucid.nn.functional.max_pool3d(input_: Tensor, kernel_size: int | tuple[int, int, int] = 1, stride: int | tuple[int, int, int] = 1, padding: int | tuple[int, int, int] = 0) → Tensor¶

The max_pool3d function performs a three-dimensional max pooling operation on the input tensor.

Function Signature¶

def max_pool3d(
    input_: Tensor,
    kernel_size: int | tuple[int, int, int] = 1,
    stride: int | tuple[int, int, int] = 1,
    padding: int | tuple[int, int, int] = 0,
) -> Tensor

Parameters¶

input_ (Tensor):
The input tensor of shape (N, C, D, H, W), where N is the batch size, C is the channels, D is the depth, H is the height, and W is the width.
kernel_size (int | tuple[int, int, int], optional):
The size of the pooling window. Default: 1.
stride (int | tuple[int, int, int], optional):
The stride of the pooling operation. Default: 1.
padding (int | tuple[int, int, int], optional):
The amount of zero-padding added to all sides of the input. Default: 0.

Returns¶

Tensor:
The result of the 3D max pooling operation, with shape (N, C, D_out, H_out, W_out), where:

\[ \begin{align}\begin{aligned}D_{out} = \frac{D + 2 \cdot \text{padding}[0] - \text{kernel depth}}{\text{stride}[0]} + 1\\H_{out} = \frac{H + 2 \cdot \text{padding}[1] - \text{kernel height}}{\text{stride}[1]} + 1\\W_{out} = \frac{W + 2 \cdot \text{padding}[2] - \text{kernel width}}{\text{stride}[2]} + 1\end{aligned}\end{align} \]

Examples¶

Performing a simple 3D max pooling:

>>> import lucid.nn.functional as F
>>> input_ = Tensor([[[[[1.0, 2.0], [3.0, 4.0]]]]])  # Shape: (1, 1, 1, 2, 2)
>>> out = F.max_pool3d(input_, kernel_size=1, stride=1, padding=0)
>>> print(out)
Tensor([[[[[1.0, 2.0], [3.0, 4.0]]]]])