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

• Operator function: performs backward propagation of the aclnnNsaCompress operator.

• Formulas: The forward propagation formulas for selected attention are as follows:

  $$\text{dw} = \text{dk\_cmp} \cdot K^\top$$ $$\text{dk} = W^\top \cdot \text{dk\_cmp}$$

[object Object]

Each operator has calls. First, [object Object] is called to obtain the workspace size required for computation and the executor that contains the operator computation process. Then, [object Object] 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 validation. The following error codes may be returned.

  [object Object]

[object Object]

• Parameters

  [object Object]

• Returns

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

[object Object]

• Deterministic computation:
  • [object Object] defaults to a deterministic implementation.
• compressBlockSize and compressStride must be integer multiples of 16 and compressBlockSize > compressStride.

[object Object]

The following single-aclnn-operator calling example (based on Atlas A2 training products) is for reference only. For details, see .