Variable

English | 简体中文

Product Support Status

Product	Supported
Ascend 950PR&950DT products	√
Atlas A3 products	√
Atlas A2 products	√
Atlas 200I/500 A2 inference products	√
Atlas inference products	√
Atlas training products	√

Function Description

Creates a variable tensor. The Variable operator assigns a value through the input x, and records the value of the variable tensor internally. The output y is the created variable tensor. The InferShape function returns success but does not explicitly set the output description, which is inferred by the framework by default (the output follows the input).

Parameter Description

Parameter Name	Input/Output/Attribute	Description	Data Type	Data Format
x	Input	The input tensor used to assign a value to the variable tensor. The caller does not need to pass the value of the variable tensor.	TensorType::ALL()	ND
index	Attribute	The index of the input tensor.	Int, with the default value of 0	-
value	Attribute	Used to pass and record the value of the variable tensor.	Tensor, with the default value of Tensor()	-
container	Attribute	The container of the variable tensor.	String, with the default value of ""	-
shared_name	Attribute	The shared name of the variable tensor.	String, with the default value of ""	-
y	Output	The created variable tensor.	TensorType::ALL()	ND

Constraint Description

None

Invocation Description

The Variable operator is a basic graph engine operator. It is usually used as a variable node in a graph, and is not invoked separately.

Invocation Method	Invocation Sample	Description
Graph-mode invocation	test_geir_variable	Invokes the Variable operator by building a graph through the operator IR.