Softmax Prototype and Function List¶

Description¶

This kernel performs Softmax activation function that is a generalization of the logistic function that transforms the input vector according to the following formula:

\[y_{i} = \frac{e^{x_{i}}}{\sum_{j}^{}e^{x_{j}}}\]

Where:

\(x_{i}\) – \(i_{\text{th}}\) value in input data subset

\(x_{j}\) – \(j_{\text{th}}\) value in the same input data subset

\(y_{i}\) – \(i_{\text{th}}\) value in output data subset

The softmax function might be applied to the whole tensor, or along a specific axis. In the first case, all the input values are involved in the calculation of each output value. If an axis is specified, then the softmax function is applied to each slice along the specific axis independently.

This kernel uses a look-up table (LUTs) to perform data transformation. See Look-Up Tables (LUT) Manipulation Prototypes and Function List section and the pseudo-code sample for more details on LUT structure preparation. Use the following functions for the purpose:

mli_krn_softmax_get_lut_size

mli_krn_softmax_create_lut

Functions¶

Kernels which implement softmax functions have the following prototype:

mli_status mli_krn_softmax_<data_format>(
   const mli_tensor *in,
   const mli_lut *lut,
   const mli_softmax_cfg *cfg,
  mli_tensor *out);

where data_format is one of the data formats listed in Table MLI Data Formats and the function parameters are shown in the following table:

Softmax Function Parameters¶
Parameter	Type	Description
`in`	`mli_tensor *`	[IN] Pointer to constant input tensor.
`lut`	`mli_lut *`	[IN] Pointer to a valid LUT table structure prepared for softmax activation.
`cfg`	`mli_softmax_cfg *`	[IN] Pointer to softmax parameters structure.
`out`	`mli_tensor *`	[IN \| OUT] Pointer to output tensor. Result is stored here

mli_softmax_cfg is defined as:

typedef mli_prelu_cfg mli_softmax_cfg;

See Table mli_prelu_cfg Structure Field Description for more details.

List of Available Softmax Functions¶
Function Name	Details
`mli_krn_softmax_sa8`	All tensors data format: sa8
`mli_krn_softmax_fx16`	All tensors data format: fx16

Conditions¶

Ensure that you satisfy the following general conditions before calling the function:

in and out tensors must be valid (see mli_tensor Structure Field Descriptions) and satisfy data requirements of the used version of the kernel.

in and out tensors must be of the same shape

lut structure must be valid and prepared for the softmax activation function (see Look-Up Tables (LUT) Manipulation Prototypes and Function List).

mem_stride of the innermost dimension must be equal to 1 for all the tensors.

axis parameter of cfg structure can be negative and must be less than in tensor rank.

For sa8 versions of kernel, in addition to general conditions, ensure that you satisfy the following quantization conditions before calling the function:

in tensors must be quantized on the tensor level. This implies that the tensor contains a single scale factor and a single zero offset.

Zero offset of in tensor must be within [-128, 127] range.

Ensure that you satisfy the platform-specific conditions in addition to those listed above (see the Platform Specific Details chapter).

Result¶

These functions modify:

Memory pointed by out.data.mem field.

el_params field of out tensor.

The range of this function is (0, 1). Depending on the data type, quantization parameters of the output tensor are configured in the following way:

fx16

out.el_params.fx.frac_bits is set to 15. Hence, the maximum representable value of softmax is equivalent to 0.999969482421875 (not 1.0).

sa8

out.el_params.sa.zero_point.mem.i16 is set to -128

out.el_params.sa.scale.mem.i16 is set to 1

out.el_params.sa.scale_frac_bits.mem.i8 is set to 8

The kernel supports in-place computation. It means that out and in tensor structures can point to the same memory with the same memory strides but without shift. It can affect performance for some platforms.

Warning

Only an exact overlap of starting address and memory stride of the in and out tensors is acceptable. Partial overlaps result in undefined behavior.

Depending on the debug level (see section Error Codes) this function performs a parameter check and returns the result as an mli_status code as described in section Kernel Specific Configuration Structures.