AIfES 2  2.0.0
ailayer_elu_default.h File Reference

Default implementation of the ELU layer . More...

Go to the source code of this file.

Data Structures

struct  ailayer_elu_f32
 Data-type specific ELU layer struct for F32 . More...
 
struct  ailayer_elu_q31
 Data-type specific ELU layer struct for Q31 . More...
 
struct  ailayer_elu_q7
 Data-type specific ELU layer struct for Q7 . More...
 

Typedefs

typedef struct ailayer_elu_f32 ailayer_elu_f32_t
 
typedef struct ailayer_elu_q31 ailayer_elu_q31_t
 
typedef struct ailayer_elu_q7 ailayer_elu_q7_t
 

Functions

ailayer_tailayer_elu_f32_default (ailayer_elu_f32_t *layer, ailayer_t *input_layer)
 Initializes and connect an ELU layer with the F32 default implementation. More...
 
ailayer_tailayer_elu_q31_default (ailayer_elu_q31_t *layer, ailayer_t *input_layer)
 Initializes and connect a ELU layer with the Q31 default implementation. More...
 
ailayer_tailayer_elu_q7_default (ailayer_elu_q7_t *layer, ailayer_t *input_layer)
 Initializes and connect a ELU layer with the Q7 default implementation. More...
 
void ailayer_elu_calc_result_tensor_params_q31_default (ailayer_t *self)
 Calculate and set the quantization parameters for the result tensor of the ELU Q31 default implementation. More...
 
void ailayer_elu_calc_result_tensor_params_q7_default (ailayer_t *self)
 Calculate and set the quantization parameters for the result tensor of the ELU Q7 default implementation. More...
 

Detailed Description

Default implementation of the ELU layer .

Version
2.2.0

Hardware independent implementations of the ELU layer in F32 data-type. For more information about the ELU layer refer to ailayer_elu.h.

Function Documentation

◆ ailayer_elu_calc_result_tensor_params_q31_default()

void ailayer_elu_calc_result_tensor_params_q31_default ( ailayer_t self)

Calculate and set the quantization parameters for the result tensor of the ELU Q31 default implementation.

Implementation of ailayer.calc_result_tensor_params.

The quantization parameters are pre-defined and the same as in the corresponding math function aimath_q31_default_elu().

Parameters
*selfThe layer structure

◆ ailayer_elu_calc_result_tensor_params_q7_default()

void ailayer_elu_calc_result_tensor_params_q7_default ( ailayer_t self)

Calculate and set the quantization parameters for the result tensor of the ELU Q7 default implementation.

Implementation of ailayer.calc_result_tensor_params.

The quantization parameters are pre-defined and the same as in the corresponding math function aimath_q7_default_elu().

Parameters
*selfThe layer structure

◆ ailayer_elu_f32_default()

ailayer_t* ailayer_elu_f32_default ( ailayer_elu_f32_t layer,
ailayer_t input_layer 
)

Initializes and connect an ELU layer with the F32 default implementation.

Example: Create the layer structure:
In C:

ailayer_elu_f32_t elu_layer = {
.alpha = 1.0f
};
Data-type specific ELU layer struct for F32 .
Definition: ailayer_elu_default.h:49
aiscalar_f32_t alpha
Data-type specific parameter used to calculate ELU function for input values < 0.
Definition: ailayer_elu_default.h:51

In C, C++ and on Arduino:

ailayer_elu_f32_t elu_layer = AILAYER_ELU_F32_A(1.0f);

Example: Initialize and connect the layer:

x = ailayer_elu_f32_default(&elu_layer, x);
ailayer_t * ailayer_elu_f32_default(ailayer_elu_f32_t *layer, ailayer_t *input_layer)
Initializes and connect an ELU layer with the F32 default implementation.
Parameters
*layerThe layer structure to initialize.
*input_layerThe prior layer.
Returns
The (successfully) initialized layer structure.

◆ ailayer_elu_q31_default()

ailayer_t* ailayer_elu_q31_default ( ailayer_elu_q31_t layer,
ailayer_t input_layer 
)

Initializes and connect a ELU layer with the Q31 default implementation.

The quantization parameters of the result tensor are automatically set to {shift = input_layer.result.shift, zero_point = input_layer.result.zero_point} because the output values are in the interval (max(-alpha, min(input_layer.result)), max(input_layer.result)].

Example: Create the layer structure:
In C:

ailayer_elu_q31_t elu_layer = {
.alpha = AISCALAR_Q31(1.0f, 2, 0)
};
Data-type specific ELU layer struct for Q31 .
Definition: ailayer_elu_default.h:58
aiscalar_q31_t alpha
Data-type specific parameter used to calculate ELU function for input values < 0.
Definition: ailayer_elu_default.h:60

In C, C++ and on Arduino:

ailayer_elu_q31_t elu_layer = AILAYER_ELU_Q31_M(AISCALAR_Q31(1.0f, 2, 0));

Example: Initialize and connect the layer:

x = ailayer_elu_q31_default(&elu_layer, x);
ailayer_t * ailayer_elu_q31_default(ailayer_elu_q31_t *layer, ailayer_t *input_layer)
Initializes and connect a ELU layer with the Q31 default implementation.
Parameters
*layerThe layer structure to initialize.
*input_layerThe prior layer.
Returns
The (successfully) initialized layer structure.

◆ ailayer_elu_q7_default()

ailayer_t* ailayer_elu_q7_default ( ailayer_elu_q7_t layer,
ailayer_t input_layer 
)

Initializes and connect a ELU layer with the Q7 default implementation.

The quantization parameters of the result tensor are automatically set to {shift = input_layer.result.shift, zero_point = input_layer.result.zero_point} because the output values are in the interval (max(-alpha, min(input_layer.result)), max(input_layer.result)].

Example: Create the layer structure:
In C:

ailayer_elu_q7_t elu_layer = {
.alpha = AISCALAR_Q7(1.0f, 2, 0)
};
Data-type specific ELU layer struct for Q7 .
Definition: ailayer_elu_default.h:67
aiscalar_q7_t alpha
Data-type specific parameter used to calculate ELU function for input values < 0.
Definition: ailayer_elu_default.h:69

In C, C++ and on Arduino:

ailayer_elu_q7_t elu_layer = AILAYER_ELU_Q7_M(AISCALAR_Q7(1.0f, 2, 0));

Example: Initialize and connect the layer:

x = ailayer_elu_q7_default(&elu_layer, x);
ailayer_t * ailayer_elu_q7_default(ailayer_elu_q7_t *layer, ailayer_t *input_layer)
Initializes and connect a ELU layer with the Q7 default implementation.
Parameters
*layerThe layer structure to initialize.
*input_layerThe prior layer.
Returns
The (successfully) initialized layer structure.