AdaptiveMaxPool1d

class torch.nn.AdaptiveMaxPool1d(output_size, return_indices=False)

Applies a 1D adaptive max pooling over an input signal composed of several input planes.

The output size is $L_{out}$, for any input size. The number of output features is equal to the number of input planes.

Each output element is the maximum over a region of the input whose bounds are derived from the input and output sizes:

$$\text{start}(k) = \left\lfloor \frac{k \times L_{in}}{L_{out}} \right\rfloor \qquad \text{end}(k) = \left\lceil \frac{(k + 1) \times L_{in}}{L_{out}} \right\rceil$$

$$out(N_i, C_j, k) = \max_{\text{start}(k) \le l < \text{end}(k)} input(N_i, C_j, l)$$

When $L_{in}$ is not divisible by $L_{out}$, these regions may have different sizes and may overlap. For example, an input of size 5 pooled to size 3 uses the regions $[0, 2)$, $[1, 4)$ and $[3, 5)$.

Parameters:

• output_size (int | tuple[int]) – the target output size $L_{out}$.

• return_indices (bool) – if True, will return the indices along with the outputs. Useful to pass to nn.MaxUnpool1d. Default: False

Shape:

• Input: $(N, C, L_{in})$ or $(C, L_{in})$.

• Output: $(N, C, L_{out})$ or $(C, L_{out})$, where $L_{out}=\text{output\_size}$.

Examples

>>> # target output size of 5
>>> m = nn.AdaptiveMaxPool1d(5)
>>> input = torch.randn(1, 64, 8)
>>> output = m(input)

forward(input)

Runs the forward pass.