FilesLast commitLast update
2 months ago
3 months ago
18 days ago
3 days ago
4 months ago
18 days ago
3 days ago
1 month ago
17 days ago
README

EluGrad

English | 简体中文

Applicable Products

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

Function

  • Operator function: Compute the backpropagation gradient corresponding to the Elu operation.

  • Formula:

    y=(min(activations,0)+1)∗gradsy = (min(activations,0) + 1)* grads

Parameters

Parameter name Input/Output/Attribute Description Data type Data format
grads Input Backpropagation gradient corresponding to the Elu operation FLOAT, FLOAT16, BFLOAT16 ND
activations Input Has the same type, format, and shape as "grads". It corresponds to the output value of the Elu operation. FLOAT, FLOAT16, BFLOAT16 ND
y Output Output tensor in the formula Same as grads ND

Constraints

None

Invocation Description

Invocation method Sample code Description
Graph mode test_geir_elu_backward.cpp Invoke the EluGrad operator through graph construction using the operator IR.