Max

Max  // Max reduction

Description

Reduces the tensor, eliminating the specified reduction axes by taking the maximum element.

Inputs

Name

Element Type

Shape

arg

Any

\((d_1,\dots,d_n)~(n \geq 0)\)

Attributes

Name

Description

reduction_axes

The axis positions (0-based) on which to calculate the max

Outputs

Name

Element Type

Shape

output

Same as arg

\((d_i:i\not\in \mathtt{reduction_axes})\)

C++ Interface

class Max : public ngraph::op::util::ArithmeticReduction

Max-reduction operation.