$OpDocStatisticsMoments
$OpDocStatisticsMoments
Input tensor.
Axes along which to compute the mean and variance.
Optional tensor of positive weights that can be broadcast with input
, to weigh the samples.
Defaults to null
, meaning that equal weighting is used (i.e., all samples have weight equal to
1
).
If true
, retain the reduced axes.
Name for the created op.
Tuple containing the created op outputs: (i) the mean tensor, and (ii) the variance tensor.
$OpDocStatisticsMomentsFromSufficientStatistics
$OpDocStatisticsMomentsFromSufficientStatistics
Total number of elements over which the provided sufficient statistics were computed.
Mean sufficient statistics: the (possibly shifted) sum of the elements.
Variance sufficient statistics: the (possibly shifted) sum of squares of the elements.
The shift by which the mean must be corrected, or null
if no shift was used.
Name for the created op.
Tuple containing the created op outputs: (i) the mean tensor, and (ii) the variance tensor.
$OpDocStatisticsSufficientStatistics
$OpDocStatisticsSufficientStatistics
Input tensor.
Tensor containing the axes along which to compute the mean and variance.
Optional tensor containing the value by which to shift the data for numerical stability.
Defaults to null
, meaning that no shift needs to be performed. A shift close to the true mean
provides the most numerically stable results.
If true
, retain the reduced axes.
Name for the created op.
Tuple containing the following created op outputs:
null
if no shift was used.