RefNextIteration
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
Makes the input tensor available in the next iteration of a loop. As a loop control flow operator, the RefNextIteration operator passes the input tensor x, which can be overwritten by references, to the next iteration, and is used to pass variable states between iterations in a while loop. The output tensor y has the same shape and data type as the input tensor x.
Parameter Description
| Parameter Name |
Input/Output/Attribute |
Description |
Data Type |
Data Format |
| x |
Input |
Input tensor. A tensor that can be overwritten by references and needs to be available in the next iteration. |
float16, float32, float64, int8, int16, int32, int64, uint8, uint16, uint32, uint64, bool |
ND |
| y |
Output |
Tensor that has the same shape and data type as the input tensor x. |
float16, float32, float64, int8, int16, int32, int64, uint8, uint16, uint32, uint64, bool |
ND |
Constraint Description
- The data type and shape of the output tensor y must be the same as those of the input tensor x.
Invocation Description
The RefNextIteration operator is a control flow operator. It usually works with the Enter and Exit operators to create loops in a graph, and is used to pass variable states between iterations. It is not invoked separately.