FilesLast commitLast update
7 days ago
1 month ago
7 days ago
22 days ago
2 days ago
9 months ago
21 days ago
2 days ago
29 days ago
8 days ago
README

ConfusionTransposeD

English | 简体中文

Product Support Status

Product Supported
Ascend 950PR&950DT products √

Function Description

  • Operator function: fuses the reshape and transpose operations.

Parameter Description

Parameter Name Input/Output/Attribute Description Data Type Data Format
x Input Input tensor. INT8, INT16, INT32, INT64, UINT8, UINT16, UINT32, UINT64, FLOAT16, FLOAT, BFLOAT16 ND
perm Input Index of the axis before transposition corresponding to each axis after transposition. INT64 ND
shape Input Shape after reshape. INT64 ND
transposeFirst Input Determines whether the transpose operation is performed first. BOOL ND
y Output Output tensor. Same as the input x. ND

Constraint Description

-Elements in the perm tensor must be unique and within the range [0, number of dimensions of perm - 1]. -When transposeFirst is True, the length of perm must be the same as the length of the shape of x, that is, len(perm)=len(x_shape); when transposeFirst is False, the length of perm must be the same as the length of the attribute input shape, that is, len(perm)=len(shape). -The product of all dimensions in shape must be equal to the total number of elements in the input tensor x.

Invocation Description

Invocation Method Invocation Sample Description
aclnn invocation test_aclnn_confusion_transpose.cpp Invokes the ConfusionTransposeD operator through the aclnnConfusionTranspose API.