nvidia.dali.fn.reductions.rms#

nvidia.dali.fn.reductions.rms(
__input,
/,
*,
axes=None,
axis_names=None,
bytes_per_sample_hint=[0],
dtype=None,
keep_dims=False,
preserve=False,
seed=-1,
device=None,
name=None,
)#

Gets root mean square of elements along provided axes.

Supported backends

  • ‘cpu’

  • ‘gpu’

Parameters:

__input (TensorList) – Input to the operator.

Keyword Arguments:

axes (int or list of int, optional) –

Axis or axes along which reduction is performed.

Accepted range is [-ndim, ndim-1]. Negative indices are counted from the back.

Not providing any axis results in reduction of all elements.

:keyword axis_names : layout str, optional: Name(s) of the axis or axes along which the reduction is performed.

The input layout is used to translate the axis names to axis indices, for example axis_names="HW" with input layout “FHWC” is equivalent to specifying axes=[1,2]. This argument cannot be used together with axes.

Keyword Arguments:

  • bytes_per_sample_hint (int or list of int, optional, default = [0]) –

    Output size hint, in bytes per sample.

    If specified, the operator’s outputs residing in GPU or page-locked host memory will be preallocated to accommodate a batch of samples of this size.

  • dtype (nvidia.dali.types.DALIDataType, optional) – Output data type. This type is used to accumulate the result.

  • keep_dims (bool, optional, default = False) – If True, maintains original input dimensions.

  • preserve (bool, optional, default = False) – Prevents the operator from being removed from the graph even if its outputs are not used.

  • seed (int, optional, default = -1) –

    Random seed.

    If not provided, it will be populated based on the global seed of the pipeline.

On this page