Gitiles
Code Review Sign In
review.mlplatform.org / ml / armnn / 36e6eaecf616f08f4d717d56e6a4ee0400d96829 / . / docs / 05_02_deserializer_serializer.dox
blob: ab821b8d66e73180cedb78906732dc4c441e8912 [file] [log] [blame]
Tracy Narine944fb502023-07-04 15:08:57 +0100[diff] [blame]1/// Copyright (c) 2021-2023 ARM Limited and Contributors. All rights reserved.
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]2///
3/// SPDX-License-Identifier: MIT
4///
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]5
6namespace armnn
7{
8/**
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]9@page serializer Serializer
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]10@tableofcontents
11
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]12The `armnnSerializer` is a library for serializing an Arm NN network to a stream.
13
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]14@section serializersupport Supported Layers
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]15
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]16This reference guide provides a list of layers which can be serialized by the Arm NN SDK.
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]17
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]18@subsection serializersupportflully Fully supported
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]19
20The Arm NN SDK Serializer currently supports the following layers:
21
22- Activation
23- Addition
24- ArgMinMax
Samuel Yapa04f4a12022-08-19 11:14:38 +0100[diff] [blame]25- BatchMatMul
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]26- BatchToSpaceNd
27- BatchNormalization
mathad01b392e982021-04-07 12:07:30 +0100[diff] [blame]28- Cast
Simon Obute51f67772021-09-03 15:50:13 +0100[diff] [blame]29- ChannelShuffle
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]30- Comparison
31- Concat
32- Constant
33- Convolution2d
Matthew Sloyanb63a3112021-09-08 13:05:51 +0100[diff] [blame]34- Convolution3d
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]35- DepthToSpace
36- DepthwiseConvolution2d
37- Dequantize
38- DetectionPostProcess
39- Division
40- ElementwiseUnary
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]41- Fill
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]42- Floor
43- FullyConnected
44- Gather
Teresa Charlin6966bfa2022-04-25 17:14:50 +0100[diff] [blame]45- GatherNd
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]46- Input
47- InstanceNormalization
48- L2Normalization
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]49- Logical
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]50- LogSoftmax
51- Lstm
52- Maximum
53- Mean
54- Merge
55- Minimum
56- Multiplication
57- Normalization
58- Output
Matthew Sloyan2e5d0b22021-10-21 14:05:31 +0100[diff] [blame]59- Pad (Constant, Symmetric, Reflect)
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]60- Permute
61- Pooling2d
Tamas Nyirid998a1c2021-11-05 14:55:33 +0000[diff] [blame]62- Pooling3d
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]63- Prelu
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]64- QLstm
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]65- Quantize
66- QuantizedLstm
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]67- Rank
Jan Eilers1121c6b2021-02-19 09:38:03 +0000[diff] [blame]68- Reduce
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]69- Reshape
70- Resize
Tracy Narine944fb502023-07-04 15:08:57 +0100[diff] [blame]71- ReverseV2
Keith Davis3ae3f972021-05-21 16:33:48 +0100[diff] [blame]72- Shape
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]73- Slice
74- Softmax
75- SpaceToBatchNd
76- SpaceToDepth
77- Splitter
78- Stack
79- StandIn
80- StridedSlice
81- Subtraction
82- Switch
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]83- Transpose
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]84- TransposeConvolution2d
Narumol Prangnawarata0162e12021-07-23 14:47:49 +0100[diff] [blame]85- UnidirectionalSequenceLstm
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]86
87More machine learning layers will be supported in future releases.
88
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]89@subsection serializersupportdeprecated Deprecated layers
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]90
91Some 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:
92
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]93- Abs will deserialize as ElementwiseUnary
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]94- Equal will deserialize as Comparison
95- Greater will deserialize as Comparison
96- Merger will deserialize as Concat
97- ResizeBilinear will deserialize as Resize
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]98- Rsqrt will deserialize as ElementwiseUnary
99<br/><br/><br/><br/>
100
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]101@page deserializer Deserializer
102@tableofcontents
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]103
104The `armnnDeserializer` is a library for loading neural networks defined by Arm NN FlatBuffers files
105into the Arm NN runtime.
106
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]107@section deserializersupport Supported Layers
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]108
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]109This reference guide provides a list of layers which can be deserialized by the Arm NN SDK.
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]110
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]111@subsection deserializersupportfully Fully supported
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]112
113The Arm NN SDK Deserialize parser currently supports the following layers:
114
115- Abs
116- Activation
117- Addition
118- ArgMinMax
Samuel Yapa04f4a12022-08-19 11:14:38 +0100[diff] [blame]119- BatchMatMul
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]120- BatchToSpaceNd
121- BatchNormalization
mathad01b392e982021-04-07 12:07:30 +0100[diff] [blame]122- Cast
Simon Obute51f67772021-09-03 15:50:13 +0100[diff] [blame]123- ChannelShuffle
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]124- Concat
125- Comparison
126- Constant
127- Convolution2d
128- DepthToSpace
129- DepthwiseConvolution2d
130- Dequantize
131- DetectionPostProcess
132- Division
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]133- ElementwiseUnary
134- Fill
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]135- Floor
136- FullyConnected
137- Gather
Teresa Charlin6966bfa2022-04-25 17:14:50 +0100[diff] [blame]138- GatherNd
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]139- Input
140- InstanceNormalization
141- L2Normalization
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]142- Logical
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]143- LogSoftmax
144- Lstm
145- Maximum
146- Mean
147- Merge
148- Minimum
149- Multiplication
150- Normalization
151- Output
152- Pad
153- Permute
154- Pooling2d
Teresa Charlin6966bfa2022-04-25 17:14:50 +0100[diff] [blame]155- Pooling3d
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]156- Prelu
157- Quantize
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]158- QLstm
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]159- QuantizedLstm
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]160- Rank
Jan Eilers1121c6b2021-02-19 09:38:03 +0000[diff] [blame]161- Reduce
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]162- Reshape
James Ward590f3f72020-12-02 10:20:48 +0000[diff] [blame]163- Resize
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]164- ResizeBilinear
Tracy Narine944fb502023-07-04 15:08:57 +0100[diff] [blame]165- ReverseV2
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]166- Slice
167- Softmax
168- SpaceToBatchNd
169- SpaceToDepth
170- Splitter
171- Stack
172- StandIn
173- StridedSlice
174- Subtraction
175- Switch
Jan Eilers1121c6b2021-02-19 09:38:03 +0000[diff] [blame]176- Transpose
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]177- TransposeConvolution2d
Narumol Prangnawarata0162e12021-07-23 14:47:49 +0100[diff] [blame]178- UnidirectionalSequenceLstm
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]179
180More machine learning layers will be supported in future releases.
181
Jan Eilers1ed388f2021-01-28 14:45:09 +0000[diff] [blame]182@subsection deserializersupportdeprecated Deprecated layers
Ryan OSheaf3a43232020-02-12 16:15:27 +0000[diff] [blame]183
184Some 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:
185
186- Equal will deserialize as Comparison
187- Merger will deserialize as Concat
188- Greater will deserialize as Comparison
189- ResizeBilinear will deserialize as Resize
190
191**/
Tracy Narine944fb502023-07-04 15:08:57 +0100[diff] [blame]192}