From a983e4699082a0b1ef685bab7354f2ad9cd37a44 Mon Sep 17 00:00:00 2001 From: Colm Donelan Date: Wed, 20 May 2020 16:12:19 +0100 Subject: Updating Doxygen documentation for 20.05 release. Change-Id: I4d624343ed5fd6ae269c3d53532903084508fd14 Signed-off-by: Colm Donelan --- 20.05/serializers.xhtml | 258 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 258 insertions(+) create mode 100644 20.05/serializers.xhtml (limited to '20.05/serializers.xhtml') diff --git a/20.05/serializers.xhtml b/20.05/serializers.xhtml new file mode 100644 index 0000000000..9704285c1a --- /dev/null +++ b/20.05/serializers.xhtml @@ -0,0 +1,258 @@ + + + + + + + + + + + + + +ArmNN: The ArmNN Serializer and Deserializer + + + + + + + + + + + + + + + + +
+
+ + + + ArmNN + + + +
+
+  20.05 +
+
+
+ + + + + + + +
+
+ +
+
+
+ +
+ +
+
+ + +
+ +
+ +
+
+
The ArmNN Serializer and Deserializer
+
+
+

Table of Contents

+ +
+

+The ArmNN Serializer

+

The armnnSerializer is a library for serializing an Arm NN network to a stream.

+

The layers that ArmNN SDK Serializer currently supports.

+

This reference guide provides a list of layers which can be serialized currently by the Arm NN SDK.

+

Fully supported

+

The Arm NN SDK Serializer currently supports the following layers:

+
    +
  • Activation
    • +
  • Addition
    • +
  • ArgMinMax
    • +
  • BatchToSpaceNd
    • +
  • BatchNormalization
    • +
  • Comparison
    • +
  • Concat
    • +
  • Constant
    • +
  • Convolution2d
    • +
  • DepthToSpace
    • +
  • DepthwiseConvolution2d
    • +
  • Dequantize
    • +
  • DetectionPostProcess
    • +
  • Division
    • +
  • ElementwiseUnary
    • +
  • Floor
    • +
  • FullyConnected
    • +
  • Gather
    • +
  • Input
    • +
  • InstanceNormalization
    • +
  • L2Normalization
    • +
  • LogSoftmax
    • +
  • Lstm
    • +
  • Maximum
    • +
  • Mean
    • +
  • Merge
    • +
  • Minimum
    • +
  • Multiplication
    • +
  • Normalization
    • +
  • Output
    • +
  • Pad
    • +
  • Permute
    • +
  • Pooling2d
    • +
  • Prelu
    • +
  • Quantize
    • +
  • QuantizedLstm
    • +
  • Reshape
    • +
  • Resize
    • +
  • Slice
    • +
  • Softmax
    • +
  • SpaceToBatchNd
    • +
  • SpaceToDepth
    • +
  • Splitter
    • +
  • Stack
    • +
  • StandIn
    • +
  • StridedSlice
    • +
  • Subtraction
    • +
  • Switch
    • +
  • TransposeConvolution2d
    • +
+

More machine learning layers will be supported in future releases.

+

Deprecated layers

+

Some 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:

+
    +
  • Equal will deserialize as Comparison
    • +
  • Merger will deserialize as Concat
    • +
  • Greater will deserialize as Comparison
    • +
  • ResizeBilinear will deserialize as Resize
    • +
  • Abs will deserialize as ElementwiseUnary
    • +
  • Rsqrt will deserialize as ElementwiseUnary
    +
    +
    +
    +
    • +
+

+The ArmNN Deserializer

+

The armnnDeserializer is a library for loading neural networks defined by Arm NN FlatBuffers files into the Arm NN runtime.

+

The layers that ArmNN SDK Deserializer currently supports.

+

This reference guide provides a list of layers which can be deserialized currently by the Arm NN SDK.

+

Fully supported

+

The Arm NN SDK Deserialize parser currently supports the following layers:

+
    +
  • Abs
    • +
  • Activation
    • +
  • Addition
    • +
  • ArgMinMax
    • +
  • BatchToSpaceNd
    • +
  • BatchNormalization
    • +
  • Concat
    • +
  • Comparison
    • +
  • Constant
    • +
  • Convolution2d
    • +
  • DepthToSpace
    • +
  • DepthwiseConvolution2d
    • +
  • Dequantize
    • +
  • DetectionPostProcess
    • +
  • Division
    • +
  • Floor
    • +
  • FullyConnected
    • +
  • Gather
    • +
  • Input
    • +
  • InstanceNormalization
    • +
  • L2Normalization
    • +
  • LogSoftmax
    • +
  • Lstm
    • +
  • Maximum
    • +
  • Mean
    • +
  • Merge
    • +
  • Minimum
    • +
  • Multiplication
    • +
  • Normalization
    • +
  • Output
    • +
  • Pad
    • +
  • Permute
    • +
  • Pooling2d
    • +
  • Prelu
    • +
  • Quantize
    • +
  • QuantizedLstm
    • +
  • Reshape
    • +
  • Rsqrt
    • +
  • Slice
    • +
  • Softmax
    • +
  • SpaceToBatchNd
    • +
  • SpaceToDepth
    • +
  • Splitter
    • +
  • Stack
    • +
  • StandIn
    • +
  • StridedSlice
    • +
  • Subtraction
    • +
  • Switch
    • +
  • TransposeConvolution2d
    • +
  • Resize
    • +
+

More machine learning layers will be supported in future releases.

+

Deprecated layers

+

Some 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:

+
    +
  • Equal will deserialize as Comparison
    • +
  • Merger will deserialize as Concat
    • +
  • Greater will deserialize as Comparison
    • +
  • ResizeBilinear will deserialize as Resize
    • +
+
+
+ + + + -- cgit v1.2.1