[object Object]

[object Object][object Object]undefined[object Object]

• Description: Performs backpropagation of .
• The Softmax function can be derived using the following formula: The relationship between out (input gradient value), gradOutput (output gradient of the previous layer), and output (Softmax forward output) can be expressed as follows:$$out = gradOutput \cdot output - sum(gradOutput \cdot output)\cdot output$$

[object Object]

Each operator has calls. First, aclnnSoftmaxBackwardGetWorkspaceSize is called to obtain the workspace size required for computation and the executor that contains the operator computation process. Then, aclnnSoftmaxBackward is called to perform computation.

[object Object]

• Parameters:

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

  The first-phase API implements input parameter verification. The following errors may be thrown:

  [object Object]

[object Object]

• Parameters:

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

[object Object]

• Deterministic compute:
  • aclnnSoftmaxBackward defaults to a deterministic implementation.

[object Object]

The following example is for reference only. For details, see .