xref: /aosp_15_r20/external/armnn/docs/05_02_deserializer_serializer.dox (revision 89c4ff92f2867872bb9e2354d150bf0c8c502810)
1/// Copyright (c) 2021 ARM Limited and Contributors. All rights reserved.
2///
3/// SPDX-License-Identifier: MIT
4///
5
6namespace armnn
7{
8/**
9@page serializer Serializer
10@tableofcontents
11
12The `armnnSerializer` is a library for serializing an Arm NN network to a stream.
13
14@section serializersupport Supported Layers
15
16This reference guide provides a list of layers which can be serialized by the Arm NN SDK.
17
18@subsection serializersupportflully Fully supported
19
20The Arm NN SDK Serializer currently supports the following layers:
21
22- Activation
23- Addition
24- ArgMinMax
25- BatchMatMul
26- BatchToSpaceNd
27- BatchNormalization
28- Cast
29- ChannelShuffle
30- Comparison
31- Concat
32- Constant
33- Convolution2d
34- Convolution3d
35- DepthToSpace
36- DepthwiseConvolution2d
37- Dequantize
38- DetectionPostProcess
39- Division
40- ElementwiseUnary
41- Fill
42- Floor
43- FullyConnected
44- Gather
45- GatherNd
46- Input
47- InstanceNormalization
48- L2Normalization
49- Logical
50- LogSoftmax
51- Lstm
52- Maximum
53- Mean
54- Merge
55- Minimum
56- Multiplication
57- Normalization
58- Output
59- Pad (Constant, Symmetric, Reflect)
60- Permute
61- Pooling2d
62- Pooling3d
63- Prelu
64- QLstm
65- Quantize
66- QuantizedLstm
67- Rank
68- Reduce
69- Reshape
70- Resize
71- Shape
72- Slice
73- Softmax
74- SpaceToBatchNd
75- SpaceToDepth
76- Splitter
77- Stack
78- StandIn
79- StridedSlice
80- Subtraction
81- Switch
82- Transpose
83- TransposeConvolution2d
84- UnidirectionalSequenceLstm
85
86More machine learning layers will be supported in future releases.
87
88@subsection serializersupportdeprecated Deprecated layers
89
90Some layers have been deprecated and replaced by others layers. In order to maintain backward compatibility, serializations of these deprecated layers will deserialize to the layers that have replaced them, as follows:
91
92- Abs will deserialize as ElementwiseUnary
93- Equal will deserialize as Comparison
94- Greater will deserialize as Comparison
95- Merger will deserialize as Concat
96- ResizeBilinear will deserialize as Resize
97- Rsqrt will deserialize as ElementwiseUnary
98<br/><br/><br/><br/>
99
100@page deserializer Deserializer
101@tableofcontents
102
103The `armnnDeserializer` is a library for loading neural networks defined by Arm NN FlatBuffers files
104into the Arm NN runtime.
105
106@section deserializersupport Supported Layers
107
108This reference guide provides a list of layers which can be deserialized by the Arm NN SDK.
109
110@subsection deserializersupportfully Fully supported
111
112The Arm NN SDK Deserialize parser currently supports the following layers:
113
114- Abs
115- Activation
116- Addition
117- ArgMinMax
118- BatchMatMul
119- BatchToSpaceNd
120- BatchNormalization
121- Cast
122- ChannelShuffle
123- Concat
124- Comparison
125- Constant
126- Convolution2d
127- DepthToSpace
128- DepthwiseConvolution2d
129- Dequantize
130- DetectionPostProcess
131- Division
132- ElementwiseUnary
133- Fill
134- Floor
135- FullyConnected
136- Gather
137- GatherNd
138- Input
139- InstanceNormalization
140- L2Normalization
141- Logical
142- LogSoftmax
143- Lstm
144- Maximum
145- Mean
146- Merge
147- Minimum
148- Multiplication
149- Normalization
150- Output
151- Pad
152- Permute
153- Pooling2d
154- Pooling3d
155- Prelu
156- Quantize
157- QLstm
158- QuantizedLstm
159- Rank
160- Reduce
161- Reshape
162- Resize
163- ResizeBilinear
164- Slice
165- Softmax
166- SpaceToBatchNd
167- SpaceToDepth
168- Splitter
169- Stack
170- StandIn
171- StridedSlice
172- Subtraction
173- Switch
174- Transpose
175- TransposeConvolution2d
176- UnidirectionalSequenceLstm
177
178More machine learning layers will be supported in future releases.
179
180@subsection deserializersupportdeprecated Deprecated layers
181
182Some layers have been deprecated and replaced by others layers. In order to maintain backward compatibility, serializations of these deprecated layers will deserialize to the layers that have replaced them, as follows:
183
184- Equal will deserialize as Comparison
185- Merger will deserialize as Concat
186- Greater will deserialize as Comparison
187- ResizeBilinear will deserialize as Resize
188
189**/
190}