[DOCS] Headings Hierarchy: Move Kernels and groups one level higher

Author: dzakhar

doc/documents/index.rst

@@ -15,6 +15,12 @@ embARC Machine Learning Inference Library
    overview/overview.rst
    mli_api_data/mli_api_data.rst
    mli_kernels/mli_kernels.rst
+   mli_kernels/convolution_grp.rst
+   mli_kernels/rec_fully_con_grp.rst
+   mli_kernels/pooling_grp.rst
+   mli_kernels/diverse_kernels_grp.rst
+   mli_kernels/transform_grp.rst
+   mli_kernels/elemw_grp.rst
    data_movement/data_movement.rst
    utility_functions/utility_functions.rst
    platform_specific/platform_hint_sum.rst

doc/documents/mli_kernels/conv_2d.rst

@@ -1,10 +1,10 @@
 .. _conv_2d:
 
 Convolution 2D Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+------------------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 This kernel implements a general 2D convolution operation. It applies each filter 
 of weights tensor to each framed area of the size of input tensor. 
@@ -38,7 +38,7 @@ related quantization aspects. The Number of accumulation series in terms of abov
 equal to :math:`(Hk * Wk * Ci)`.
 
 Functions
-"""""""""
+^^^^^^^^^
 
 The functions which implement 2D Convolutions have the following prototype:
 
@@ -162,7 +162,7 @@ Here is a list of all available 2D Convolution functions:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function:
 
@@ -257,7 +257,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions only modify the memory pointed by ``out.data.mem`` field. 
 It is assumed that all the other fields of ``out`` tensor are properly populated 

doc/documents/mli_kernels/conv_depthwise.rst

@@ -1,10 +1,10 @@
 .. _conv_depthwise:
 
 Depthwise Convolution Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+-------------------------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 This kernel implements a 2D depthwise convolution operation applying each filter 
 channel to each input channel separately. It applies each filter of weights tensor 
@@ -45,7 +45,7 @@ on related quantization aspects. The Number of accumulation series in terms of a
 is equal to :math:`(Hk * Wk)`.
 
 Functions
-"""""""""
+^^^^^^^^^
 
 Kernels which implement depthwise convolution have the following prototype:
 
@@ -148,7 +148,7 @@ Here is a list of all available Depth-Wise Convolution functions:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function:
 
@@ -223,7 +223,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions only modify the memory pointed by ``out.data.mem`` field. 
 It is assumed that all the other fields of ``out`` tensor are properly populated 

doc/documents/mli_kernels/conv_grp.rst

@@ -1,10 +1,10 @@
 .. _grp_conv:
 
 Group Convolution Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+---------------------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 This kernel implements a grouped 2D convolution which applies general 2D 
 convolution operation on a separate subset (groups) of inputs. In grouped 
@@ -41,7 +41,7 @@ This is a MAC-based kernel which implies accumulation. See :ref:`quant_accum_inf
 The Number of accumulation series is equal to :math:`(Hk * Wk * Cw)`.
 
 Functions
-"""""""""
+^^^^^^^^^
 
 Kernels which implement a group convolution have the following prototype:
 
@@ -152,7 +152,7 @@ Here is a list of all available Group Convolution functions:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 Ensure that you satisfy the following general conditions before calling the function:
 
 
@@ -228,7 +228,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions only modify the memory pointed by ``out.data.mem`` field. 
 It is assumed that all the other fields of ``out`` tensor are properly populated 

doc/documents/mli_kernels/conv_transp.rst

@@ -1,8 +1,8 @@
 Transpose Convolution Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+-------------------------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 This kernel implements a general 2D transposed convolution operation 
 which works by swapping the forward and backward passes of a convolution. 
@@ -55,7 +55,7 @@ This is a MAC-based kernel which implies accumulation. See :ref:`quant_accum_inf
 The Number of accumulation series is up to (:math:`Wk*Hk*Ci`).
 
 Functions
-"""""""""
+^^^^^^^^^
 
 Kernels which implement Transpose Convolutions have the following prototype:
 
@@ -170,7 +170,7 @@ The following table lists all the available Transpose Convolution functions:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function: 
 
@@ -241,7 +241,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions only modify the memory pointed by ``out.data.mem`` field. 
 It is assumed that all the other fields of ``out`` tensor are properly populated 

doc/documents/mli_kernels/convolution_grp.rst

@@ -1,7 +1,7 @@
 .. _chap_conv:
 
-Convolution Group
------------------
+Convolution Kernels Group
+=========================
 
 The Convolution Group describes operations related to convolution layers, where 
 input features are convolved with a set of trained filters. For a comprehensive guide 

doc/documents/mli_kernels/div_argmax.rst

@@ -1,16 +1,16 @@
 .. _argmax_prot:
 
 Argmax Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+----------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 This kernel returns the positions of maximum values across the whole tensor, or for each slice 
 across a dimension. 
 
 Functions
-"""""""""
+^^^^^^^^^
 
 Argmax functions have the following prototype:
 
@@ -82,7 +82,7 @@ parameters are shown in the following table:
 ..   
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function:
 
@@ -126,7 +126,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions modify:
 

doc/documents/mli_kernels/div_permute.rst

@@ -1,16 +1,16 @@
 .. _permute_prot:
 
 Permute Prototype and Function List
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+-----------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 The kernel permutes dimensions of input tensor according to provided order. In other words,
 it transposes input tensors.
 
 Functions
-"""""""""
+^^^^^^^^^
 
 The functions which implement Permute have the following prototype:
 
@@ -86,7 +86,7 @@ Here is a list of all available permute functions:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function:
 
@@ -104,7 +104,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions modify:
 

doc/documents/mli_kernels/diverse_kernels_grp.rst

@@ -1,7 +1,7 @@
 .. _chap_diverse_kern:
 
 Diverse Kernels Group 
----------------------
+=====================
 
 The Diverse Kernels Group provides operations for common machine learning processing. 
 

doc/documents/mli_kernels/elemw_grp.rst

@@ -1,10 +1,13 @@
 .. _chap_element_wise:
 
-Element-wise Group
-^^^^^^^^^^^^^^^^^^
+Element-wise Kernels Group
+==========================
+
+Element-wise Operations Prototype and Function List
+---------------------------------------------------
 
 Description
-"""""""""""
+^^^^^^^^^^^
 
 The Element-wise Group describes operations that are applied element-by-element 
 on two tensors of the same shape and return a tensor of the same shape. These kernels 
@@ -17,7 +20,7 @@ element of the other tensor.
 :math:`\text{out}_{i} = operation(\text{in}_{\text{scalar}}^{1},\ \text{in}_{i}^{2}`)
 
 Functions
-"""""""""
+^^^^^^^^^
 
 Kernels which implement Element-wise functions have the following prototype:
 
@@ -76,7 +79,7 @@ Kernels which implement Element-wise functions have the following prototype:
 ..
 
 Conditions
-""""""""""
+^^^^^^^^^^
 
 Ensure that you satisfy the following general conditions before calling the function:
 
@@ -112,7 +115,7 @@ Ensure that you satisfy the platform-specific conditions in addition to those li
 (see the :ref:`platform_spec_chptr` chapter).
 
 Result
-""""""
+^^^^^^
 
 These functions only modify the memory pointed by ``out.data.mem`` field. 
 It is assumed that all the other fields of ``out`` tensor are properly populated