NcclReduce

Stay organized with collections Save and categorize content based on your preferences.
public final class NcclReduce

Reduces `input` from `num_devices` using `reduction` to a single device.

Reduces `input` from `num_devices` using `reduction` to a single device.

The graph should be constructed so that all inputs have a valid device assignment, and the op itself is assigned one of these devices.

input: The input to the reduction. data: the value of the reduction across all `num_devices` devices. reduction: the reduction operation to perform.

Constants

String OP_NAME The name of this op, as known by TensorFlow core engine

Public Methods

Output<T>
asOutput()
Returns the symbolic handle of the tensor.
static <T extends TNumber> NcclReduce<T>
create(Scope scope, Iterable<Operand<T>> input, String reduction)
Factory method to create a class wrapping a new NcclReduce operation.
Output<T>
data()

Inherited Methods

Constants

public static final String OP_NAME

The name of this op, as known by TensorFlow core engine

Constant Value: "NcclReduce"

Public Methods

public Output<T> asOutput ()

Returns the symbolic handle of the tensor.

Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.

public static NcclReduce<T> create (Scope scope, Iterable<Operand<T>> input, String reduction)

Factory method to create a class wrapping a new NcclReduce operation.

Parameters
scope current scope
Returns
  • a new instance of NcclReduce

public Output<T> data ()