[DOCS] Grammar corrections

Author: dzakhar

doc/documents/data_movement/data_movement.rst

@@ -132,6 +132,7 @@ The size of the array is defined by ``MLI_MAX_RANK``.
 
 It is possible to combine multiple 'operations' in one move. In that the internal order of how the parameters are
 applied is relevant:
+
  - padding_pre/padding_post (using cfg->padding_pre and cfg->padding_post as described in figure :ref:`f_mli_mov_cfg_params_pad`)
  - crop  (using cfg->offset and cfg->size as described in figure :ref:`f_mli_mov_cfg_params_crop`)
  - subsampling (using cfg->sub_sample_step as described in figure :ref:`f_mli_mov_cfg_params_sub`)
@@ -193,7 +194,7 @@ following options to initialize all the fields in a consistent way:
 Some operations (like padding, concat, slice, subsample) when applied on the quantization axis will affect the
 quantization parameters (e.g. If subsampling on the quantization axis is applied, also the quantization parameters
 will be subsampled). For these cases the output tensor should contain valid pre-allocated memory buffers to get
-correct quantization parameters, if this is not done, and the pointers are initialized with ``nullptr`` or with the
+correct quantization parameters. If this is not done, and the pointers are initialized with ``nullptr`` or with the
 same pointers as the src tensor, the result is undefined.
 
     - In case of per-axis quantization of the src tensor, the axis needs to match the (permuted) quantization axis in the

doc/documents/mli_api_data/data_structures.rst

@@ -84,7 +84,7 @@ The table :ref:`mli_tnsr_struc` describes the fields in the ``mli_tensor`` struc
    |                   |                        |  - ``mem_stride[2] >= 1``                                                   |
    |                   |                        |                                                                             |
    |                   |                        | ``mli_move`` is the only function which can write the ``mem_stride`` field  |
-   |                   |                        | of the ``dst`` tensor. Other kernels don't update this Field                |
+   |                   |                        | of the ``dst`` tensor. Other kernels don't update this field                |
    +-------------------+------------------------+-----------------------------------------------------------------------------+
    | ``rank``          | ``uint32_t``           | Number of dimensions of this tensor (Must be less or equal to               |
    |                   |                        | ``MLI_MAX_RANK*``)                                                          |
@@ -106,11 +106,11 @@ The table :ref:`mli_tnsr_struc` describes the fields in the ``mli_tensor`` struc
 ..
 
 .. important::
-   ``mli_tensor`` is valid if all it's fields are populated in a non-contradictory way which implies: 
+   ``mli_tensor`` is valid if all its fields are populated in a non-contradictory way which implies: 
 
       - ``rank`` and ``shape`` fields are aligned with ``mem_strides`` field
       - ``data`` container points to a memory region of ``el_type`` elements or contains a single 
-        element itself (see :ref:`c_mli_data_container`). It’s capacity is enough to preserve data described 
+        element itself (see :ref:`c_mli_data_container`). Its capacity is enough to preserve data described 
         by ``rank``, ``shape`` and ``mem_stride`` fields.
       - ``el_params`` structure is filled properly according to ``el_type`` field and
         related quantization scheme (see :ref:`data_fmts`)
@@ -186,7 +186,7 @@ It stores pointer to data or a single value that intend to be directly used in a
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 ``mli_element_type`` enumeration defines the basic element type stored in tensor structure.
-Based on this information library functions may define sizes, algorithms for processing,
+Based on this information, library functions may define sizes, algorithms for processing,
 and other  implementation specific things. ``mli_element_type`` is defined as follows:   
 
 .. code:: c

doc/documents/mli_kernels/conv_2d.rst

@@ -167,16 +167,16 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in``, ``out``, ``weights`` and ``bias`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - Shapes of ``in``, ``out``, ``weights`` and ``bias`` tensors must be compatible,
    which implies the following requirements:
 
     - ``in`` and ``out`` are 3-dimensional tensors (rank==3). Dimensions meaning, 
-      and order (layout) is aligned with the used version of kernel.
+      and order (layout) is aligned with the specific version of kernel.
 
     - ``weights`` is a 4-dimensional tensor (rank==4). Dimensions meaning, 
-      and order (layout) is aligned with the used kernel.
+      and order (layout) is aligned with the specific kernel.
 
     - ``bias`` must be a one-dimensional tensor (rank==1). Its length must be equal to 
       :math:`Co` (output channels OR number of filters).
@@ -244,7 +244,7 @@ the following quantization conditions before calling the function:
        const float w_scale_1 = 0.00542382f;
        float bias_scale = in_w_scale * w_scale_1;
        
-       // Derive quantized bias scale and frac bits for it to use it in tensor struct.
+       // Derive quantized bias scale and frac bits for use in tensor struct.
        int exp;
        frexpf(bias_scale, &exp);
        int bias_scale_frac_bits = 15 - exp;

doc/documents/mli_kernels/conv_depthwise.rst

@@ -153,16 +153,16 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in``, ``out``, ``weights`` and ``bias`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - Shapes of ``in``, ``out``, ``weights`` and ``bias`` tensors must be compatible,
    which implies the following requirements:
 
    - ``in`` and ``out`` are 3-dimensional tensors (rank==3). Dimensions meaning, 
-     and order (layout) is aligned with the used version of kernel.
+     and order (layout) is aligned with the specific version of kernel.
 
    - ``weights`` is a 4-dimensional tensor (rank==4). Dimensions meaning, 
-     and order (layout) is aligned with the used kernel.
+     and order (layout) is aligned with the specifc kernel.
 
    - ``bias`` must be a one-dimensional tensor (rank==1). Its length must be equal to 
      :math:`Co` (output channels OR number of filters).

doc/documents/mli_kernels/conv_grp.rst

@@ -15,7 +15,7 @@ outputs are then concatenated together to give the final output.
 
 For example, in a HWCN data layout, if the ``in`` feature map is :math:`(Hi, Wi, Ci)` and 
 the ``weights`` is :math:`(Hk, Wk, Cw, Co)`, the ``output`` feature map is :math:`(Ho, Wo, Co)`
-tensor where :math:`Ci` is equalt to :math:`Cw * M` and :math:`Co` is multiple of :math:`M`.
+tensor where :math:`Ci` is equal to :math:`Cw * M` and :math:`Co` is multiple of :math:`M`.
 Also spatial dimensions :math:`H*, W*` comply with the system of equations :eq:`eq_conv2d_shapes`. 
 
 Depthwise convolution (see :ref:`conv_depthwise`) is an extreme case of group 
@@ -157,16 +157,16 @@ Ensure that you satisfy the following general conditions before calling the func
 
 
 - ``in``, ``out``, ``weights`` and ``bias`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-  and satisfy data requirements of the used version of the kernel.
+  and satisfy data requirements of the selected version of the kernel.
 
 -  Shapes of ``in``, ``out``, ``weights`` and ``bias`` tensors must be compatible,
    which implies the following requirements:
 
     - ``in`` and ``out`` are 3-dimensional tensors (rank==3). Dimensions meaning, 
-      and order (layout) is aligned with the used version of kernel.
+      and order (layout) is aligned with the specifc version of kernel.
 
     - ``weights`` is a 4-dimensional tensor (rank==4). Dimensions meaning, 
-      and order (layout) is aligned with the used kernel.
+      and order (layout) is aligned with the specific kernel.
 
     - ``bias`` must be a one-dimensional tensor (rank==1). Its length must be equal to 
       :math:`Co` (output channels OR number of filters).

doc/documents/mli_kernels/conv_transp.rst

@@ -175,16 +175,16 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function: 
 
  - ``in``, ``out``, ``weights`` and ``bias`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - Shapes of ``in``, ``out``, ``weights`` and ``bias`` tensors must be compatible,
    which implies the following requirements:
 
     - ``in`` and ``out`` are 3-dimensional tensors (rank==3). Dimensions meaning, 
-      and order (layout) is aligned with the used version of kernel.
+      and order (layout) is aligned with the specific version of kernel.
 
     - ``weights`` is a 4-dimensional tensor (rank==4). Dimensions meaning, 
-      and order (layout) is aligned with the used kernel.
+      and order (layout) is aligned with the specific kernel.
 
     - ``bias`` must be a one-dimensional tensor (rank==1). Its length must be equal to 
       :math:`Co` (output channels OR number of filters).

doc/documents/mli_kernels/div_argmax.rst

@@ -87,7 +87,7 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in`` tensor must be valid (see :ref:`mli_tnsr_struc`) 
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - ``out`` tensor must be valid (see :ref:`mli_tnsr_struc`) 
    and satisfy the following requiremens:

doc/documents/mli_kernels/div_permute.rst

@@ -91,7 +91,7 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in`` and ``out`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - Shape of ``out`` tensor must reflect already permuted  shape of ``in`` tensor.
 
@@ -100,7 +100,7 @@ Ensure that you satisfy the following general conditions before calling the func
  - ``mem_stride`` of the innermost dimension must be equal to 1 for all the tensors.
 
  - Only first N (equal to ``rank`` of ``in`` tensor) values in permutation order array are considered 
-   by kernel. All of them must be unique, nonnegative and less than the ``rank`` of the ``in`` tensor.
+   by kernel. All of them must be unique, non-negative and less than the ``rank`` of the ``in`` tensor.
 
 Result
 ^^^^^^
@@ -113,7 +113,7 @@ These functions modify:
 It is assumed that all the other fields and structures are properly populated 
 to be used in calculations and are not modified by the kernel.
 
-For **fx8** and **fx16** ``el_params`` field of ``in`` tensor is copyed to the ``out`` tensor. The same is applyed for
+For **fx8** and **fx16** ``el_params`` field of ``in`` tensor is copyed to the ``out`` tensor. The same is applied for
 **sa8** versions of kernel in case of per-tensor quantization (``in.el_params.sa.dim`` < 0)
 
 For **sa8** versions of kernel, and in case of per-axis quantization, the ``el_params`` field of the 

doc/documents/mli_kernels/elemw_grp.rst

@@ -81,7 +81,7 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in1``, ``in2`` and ``out`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the specific version of the kernel.
 
  - Shapes of ``in1``, ``in2`` and ``out`` tensors must be compatible,
    which implies the following requirements:

doc/documents/mli_kernels/pool_avg.rst

@@ -93,13 +93,13 @@ Conditions
 Ensure that you satisfy the following general conditions before calling the function:
 
  - ``in`` and ``out`` tensors must be valid (see :ref:`mli_tnsr_struc`)
-   and satisfy data requirements of the used version of the kernel.
+   and satisfy data requirements of the selected version of the kernel.
 
  - Shapes of ``in``  and  ``out`` tensors must be compatible,
    which implies the following requirements:
 
    - ``in`` and ``out`` are 3-dimensional tensors (rank==3). Dimensions meaning, 
-     and order (layout) is aligned with the used version of kernel.
+     and order (layout) is aligned with the specific version of kernel.
 
    - Channel :math:`Ci` dimension of ``in`` and ``out`` tensors must be equal.