pool¶

dragon.nn.pool(
  inputs,
  kernel_shape,
  strides,
  pads=0,
  padding='VALID',
  mode='max',
  global_pool=False,
  ceil_mode=False,
  data_format='NCHW',
  **kwargs
)[source]¶

Apply the n-dimension pooling.

• Set mode for the specific pooling type, default is maxpool.
• Use global_pool to apply the global pooling further.
• If data_format is 'NCHW', excepts input shape \((N, C, D1, D2, ...)\), and output shape is \((N, C, D1_{\text{out}}, D2_{\text{out}}, ...)\).
• If data_format is 'NHWC', excepts input shape \((N, D1, D2, ..., C)\), and output shape is \((N, D1_{\text{out}}, D2_{\text{out}}, ..., C)\).
• If padding is 'VALID', pads controls the explicit padding size. Otherwise, size are computed automatically use the given method.

Examples:

for i in range(3):
    ndim = i + 1
    x = dragon.ones((1, 2) + (2,) * ndim)
    y = dragon.nn.pool(x, kernel_shape=(2,) * ndim, strides=2)
    assert y.shape == (1, 2) + (1,) * ndim

Parameters:

• inputs (dragon.Tensor) – The input tensor.
• kernel_shape (Sequence[int], required) – The shape of pooling window.
• strides (Union[int, Sequence[int]], required) – The stride of pooling window.
• pads (Union[int, Sequence[int]], optional, default=0) – The zero padding size.
• padding (str, optional, default='VALID') – 'VALID', 'SAME', 'SAME_UPPER' or 'SAME_LOWER'.
• mode (str, optional, default='max') – 'max' or 'avg'.
• global_pool (bool, optional, default=False) – Apply the global pooling or not.
• ceil_mode (bool, optional, default=False) – Ceil or floor the boundary.
• data_format (str, optional, default='NCHW') – 'NCHW' or 'NHWC'.

Returns:

dragon.Tensor – The output tensor.