[object Object]

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

• Description:

  Multiplies the input [object Object] by the corresponding [object Object] and [object Object] based on the input index values in [object Object], and accumulates the results into the input [object Object]to produce the output.

• Formulas:

  Given an input tensor [object Object] with the last dimension length of 2d, the AddLora function performs the following operations:

  1. Rearrange [object Object] based on the index values in [object Object]. Elements of [object Object] corresponding to the same weight group are placed together.

  2. Loop through each LoRA group and perform matrix multiplication using its corresponding [object Object] element and [object Object].

     $$Z1 = x_{i} \cdot weightA[i, layerIdx, :, :]$$

  3. Multiply the resultant [object Object] with [object Object].

     $$Z2 = Z1 \cdot weightB[i, layerIdx, :, :] \times scale$$

  4. Finally, accumulate [object Object] into [object Object] to obtain the output.

     $$\text{out} = y[:, yOffset: yOffset+ySliceSize] + Z2$$

[object Object]

Each operator has calls. First, [object Object] is called to obtain the input parameters and compute the required workspace size based on the computation process. Then, [object Object] is called to perform computation.

[object Object]

• Parameters:

  [object Object]
  • [object Object]Atlas A2 training products/Atlas A2 inference products[object Object]: The data formats of [object Object]weightB[object Object] and [object Object]weightAOptional[object Object] can be ND.

• 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 computation:
  • [object Object] defaults to a deterministic implementation.

[object Object]

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