1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
|
//
// Copyright © 2019 Arm Ltd and Contributors. All rights reserved.
// SPDX-License-Identifier: MIT
//
#include "ParserFlatbuffersFixture.hpp"
TEST_SUITE("TensorflowLiteParser_Transpose")
{
struct TransposeFixture : public ParserFlatbuffersFixture
{
explicit TransposeFixture(const std::string & inputShape,
const std::string & permuteData,
const std::string & outputShape)
{
m_JsonString = R"(
{
"version": 3,
"operator_codes": [
{
"builtin_code": "TRANSPOSE",
"version": 1
}
],
"subgraphs": [
{
"tensors": [
{
"shape": )" + inputShape + R"(,
"type": "FLOAT32",
"buffer": 0,
"name": "inputTensor",
"quantization": {
"min": [
0.0
],
"max": [
255.0
],
"details_type": 0,
"quantized_dimension": 0
},
"is_variable": false
},
{
"shape": )" + outputShape + R"(,
"type": "FLOAT32",
"buffer": 1,
"name": "outputTensor",
"quantization": {
"details_type": 0,
"quantized_dimension": 0
},
"is_variable": false
})";
m_JsonString += R"(,
{
"shape": [
3
],
"type": "INT32",
"buffer": 2,
"name": "permuteTensor",
"quantization": {
"details_type": 0,
"quantized_dimension": 0
},
"is_variable": false
})";
m_JsonString += R"(],
"inputs": [
0
],
"outputs": [
1
],
"operators": [
{
"opcode_index": 0,
"inputs": [
0)";
m_JsonString += R"(,2)";
m_JsonString += R"(],
"outputs": [
1
],
"builtin_options_type": "TransposeOptions",
"builtin_options": {
},
"custom_options_format": "FLEXBUFFERS"
}
]
}
],
"description": "TOCO Converted.",
"buffers": [
{ },
{ })";
if (!permuteData.empty())
{
m_JsonString += R"(,{"data": )" + permuteData + R"( })";
}
m_JsonString += R"(
]
}
)";
Setup();
}
};
// Note that this assumes the Tensorflow permutation vector implementation as opposed to the armnn implemenation.
struct TransposeFixtureWithPermuteData : TransposeFixture
{
TransposeFixtureWithPermuteData() : TransposeFixture("[ 2, 2, 3 ]",
"[ 0, 0, 0, 0, 2, 0, 0, 0, 1, 0, 0, 0 ]",
"[ 2, 3, 2 ]") {}
};
TEST_CASE_FIXTURE(TransposeFixtureWithPermuteData, "TransposeWithPermuteData")
{
RunTest<3, armnn::DataType::Float32>(
0,
{{"inputTensor", { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 }}},
{{"outputTensor", { 1, 4, 2, 5, 3, 6, 7, 10, 8, 11, 9, 12 }}});
CHECK((m_Parser->GetNetworkOutputBindingInfo(0, "outputTensor").second.GetShape()
== armnn::TensorShape({2,3,2})));
}
// Tensorflow default permutation behavior assumes no permute argument will create permute vector [n-1...0],
// where n is the number of dimensions of the input tensor
// In this case we should get output shape 3,2,2 given default permutation vector 2,1,0
struct TransposeFixtureWithoutPermuteData : TransposeFixture
{
TransposeFixtureWithoutPermuteData() : TransposeFixture("[ 2, 2, 3 ]",
"[ 2, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0 ]",
"[ 3, 2, 2 ]") {}
};
TEST_CASE_FIXTURE(TransposeFixtureWithoutPermuteData, "TransposeWithoutPermuteDims")
{
RunTest<3, armnn::DataType::Float32>(
0,
{{"inputTensor", { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 }}},
{{"outputTensor", { 1, 7, 4, 10, 2, 8, 5, 11, 3, 9, 6, 12 }}});
CHECK((m_Parser->GetNetworkOutputBindingInfo(0, "outputTensor").second.GetShape()
== armnn::TensorShape({3,2,2})));
}
}
|
