Hugging Face's logo

camstack
/
camstack-models

TF-Keras
LiteRT
Core ML
ONNX
Model card Files
xet
 Community
1
apocaliss92 commited on
Commit
3a326e0
·
verified ·
1 Parent(s): 1b64c51

Re-convert valid OpenVINO IR for vgg-english-g2

Browse files
Files changed (1) hide show
  1. plateRecognition/vgg_english_g2/openvino/vgg_english_g2.xml +2342 -0
plateRecognition/vgg_english_g2/openvino/vgg_english_g2.xml ADDED
@@ -0,0 +1,2342 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0"?>
2
+ <net name="onnx_Frontend_IR" version="11">
3
+ <layers>
4
+ <layer id="0" name="images" type="Parameter" version="opset1">
5
+ <data shape="1,1,64,384" element_type="f32" />
6
+ <output>
7
+ <port id="0" precision="FP32" names="images">
8
+ <dim>1</dim>
9
+ <dim>1</dim>
10
+ <dim>64</dim>
11
+ <dim>384</dim>
12
+ </port>
13
+ </output>
14
+ </layer>
15
+ <layer id="1" name="Constant_275727_compressed" type="Const" version="opset1">
16
+ <data element_type="f16" shape="1, 1, 97" offset="0" size="194" />
17
+ <output>
18
+ <port id="0" precision="FP16">
19
+ <dim>1</dim>
20
+ <dim>1</dim>
21
+ <dim>97</dim>
22
+ </port>
23
+ </output>
24
+ </layer>
25
+ <layer id="2" name="Constant_275727" type="Convert" version="opset1">
26
+ <data destination_type="f32" />
27
+ <rt_info>
28
+ <attribute name="decompression" version="0" />
29
+ </rt_info>
30
+ <input>
31
+ <port id="0" precision="FP16">
32
+ <dim>1</dim>
33
+ <dim>1</dim>
34
+ <dim>97</dim>
35
+ </port>
36
+ </input>
37
+ <output>
38
+ <port id="1" precision="FP32">
39
+ <dim>1</dim>
40
+ <dim>1</dim>
41
+ <dim>97</dim>
42
+ </port>
43
+ </output>
44
+ </layer>
45
+ <layer id="3" name="Constant_275726_compressed" type="Const" version="opset1">
46
+ <data element_type="f16" shape="1, 1, 256" offset="194" size="512" />
47
+ <output>
48
+ <port id="0" precision="FP16">
49
+ <dim>1</dim>
50
+ <dim>1</dim>
51
+ <dim>256</dim>
52
+ </port>
53
+ </output>
54
+ </layer>
55
+ <layer id="4" name="Constant_275726" type="Convert" version="opset1">
56
+ <data destination_type="f32" />
57
+ <rt_info>
58
+ <attribute name="decompression" version="0" />
59
+ </rt_info>
60
+ <input>
61
+ <port id="0" precision="FP16">
62
+ <dim>1</dim>
63
+ <dim>1</dim>
64
+ <dim>256</dim>
65
+ </port>
66
+ </input>
67
+ <output>
68
+ <port id="1" precision="FP32">
69
+ <dim>1</dim>
70
+ <dim>1</dim>
71
+ <dim>256</dim>
72
+ </port>
73
+ </output>
74
+ </layer>
75
+ <layer id="5" name="Constant_275725_compressed" type="Const" version="opset1">
76
+ <data element_type="f16" shape="1, 1, 256" offset="706" size="512" />
77
+ <output>
78
+ <port id="0" precision="FP16">
79
+ <dim>1</dim>
80
+ <dim>1</dim>
81
+ <dim>256</dim>
82
+ </port>
83
+ </output>
84
+ </layer>
85
+ <layer id="6" name="Constant_275725" type="Convert" version="opset1">
86
+ <data destination_type="f32" />
87
+ <rt_info>
88
+ <attribute name="decompression" version="0" />
89
+ </rt_info>
90
+ <input>
91
+ <port id="0" precision="FP16">
92
+ <dim>1</dim>
93
+ <dim>1</dim>
94
+ <dim>256</dim>
95
+ </port>
96
+ </input>
97
+ <output>
98
+ <port id="1" precision="FP32">
99
+ <dim>1</dim>
100
+ <dim>1</dim>
101
+ <dim>256</dim>
102
+ </port>
103
+ </output>
104
+ </layer>
105
+ <layer id="7" name="FeatureExtraction.ConvNet.0.weight_compressed" type="Const" version="opset1">
106
+ <data element_type="f16" shape="32, 1, 3, 3" offset="1218" size="576" />
107
+ <output>
108
+ <port id="0" precision="FP16" names="FeatureExtraction.ConvNet.0.weight">
109
+ <dim>32</dim>
110
+ <dim>1</dim>
111
+ <dim>3</dim>
112
+ <dim>3</dim>
113
+ </port>
114
+ </output>
115
+ </layer>
116
+ <layer id="8" name="FeatureExtraction.ConvNet.0.weight" type="Convert" version="opset1">
117
+ <data destination_type="f32" />
118
+ <rt_info>
119
+ <attribute name="decompression" version="0" />
120
+ </rt_info>
121
+ <input>
122
+ <port id="0" precision="FP16">
123
+ <dim>32</dim>
124
+ <dim>1</dim>
125
+ <dim>3</dim>
126
+ <dim>3</dim>
127
+ </port>
128
+ </input>
129
+ <output>
130
+ <port id="1" precision="FP32">
131
+ <dim>32</dim>
132
+ <dim>1</dim>
133
+ <dim>3</dim>
134
+ <dim>3</dim>
135
+ </port>
136
+ </output>
137
+ </layer>
138
+ <layer id="9" name="/FeatureExtraction/ConvNet/ConvNet.0/Conv/WithoutBiases" type="Convolution" version="opset1">
139
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
140
+ <input>
141
+ <port id="0" precision="FP32">
142
+ <dim>1</dim>
143
+ <dim>1</dim>
144
+ <dim>64</dim>
145
+ <dim>384</dim>
146
+ </port>
147
+ <port id="1" precision="FP32">
148
+ <dim>32</dim>
149
+ <dim>1</dim>
150
+ <dim>3</dim>
151
+ <dim>3</dim>
152
+ </port>
153
+ </input>
154
+ <output>
155
+ <port id="2" precision="FP32">
156
+ <dim>1</dim>
157
+ <dim>32</dim>
158
+ <dim>64</dim>
159
+ <dim>384</dim>
160
+ </port>
161
+ </output>
162
+ </layer>
163
+ <layer id="10" name="Reshape_273128" type="Const" version="opset1">
164
+ <data element_type="f32" shape="1, 32, 1, 1" offset="1794" size="128" />
165
+ <output>
166
+ <port id="0" precision="FP32">
167
+ <dim>1</dim>
168
+ <dim>32</dim>
169
+ <dim>1</dim>
170
+ <dim>1</dim>
171
+ </port>
172
+ </output>
173
+ </layer>
174
+ <layer id="11" name="/FeatureExtraction/ConvNet/ConvNet.0/Conv" type="Add" version="opset1">
175
+ <data auto_broadcast="numpy" />
176
+ <input>
177
+ <port id="0" precision="FP32">
178
+ <dim>1</dim>
179
+ <dim>32</dim>
180
+ <dim>64</dim>
181
+ <dim>384</dim>
182
+ </port>
183
+ <port id="1" precision="FP32">
184
+ <dim>1</dim>
185
+ <dim>32</dim>
186
+ <dim>1</dim>
187
+ <dim>1</dim>
188
+ </port>
189
+ </input>
190
+ <output>
191
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.0/Conv_output_0">
192
+ <dim>1</dim>
193
+ <dim>32</dim>
194
+ <dim>64</dim>
195
+ <dim>384</dim>
196
+ </port>
197
+ </output>
198
+ </layer>
199
+ <layer id="12" name="/FeatureExtraction/ConvNet/ConvNet.1/Relu" type="ReLU" version="opset1">
200
+ <input>
201
+ <port id="0" precision="FP32">
202
+ <dim>1</dim>
203
+ <dim>32</dim>
204
+ <dim>64</dim>
205
+ <dim>384</dim>
206
+ </port>
207
+ </input>
208
+ <output>
209
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.1/Relu_output_0">
210
+ <dim>1</dim>
211
+ <dim>32</dim>
212
+ <dim>64</dim>
213
+ <dim>384</dim>
214
+ </port>
215
+ </output>
216
+ </layer>
217
+ <layer id="13" name="/FeatureExtraction/ConvNet/ConvNet.2/MaxPool" type="MaxPool" version="opset8">
218
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" kernel="2, 2" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
219
+ <input>
220
+ <port id="0" precision="FP32">
221
+ <dim>1</dim>
222
+ <dim>32</dim>
223
+ <dim>64</dim>
224
+ <dim>384</dim>
225
+ </port>
226
+ </input>
227
+ <output>
228
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.2/MaxPool_output_0">
229
+ <dim>1</dim>
230
+ <dim>32</dim>
231
+ <dim>32</dim>
232
+ <dim>192</dim>
233
+ </port>
234
+ <port id="2" precision="I64">
235
+ <dim>1</dim>
236
+ <dim>32</dim>
237
+ <dim>32</dim>
238
+ <dim>192</dim>
239
+ </port>
240
+ </output>
241
+ </layer>
242
+ <layer id="14" name="FeatureExtraction.ConvNet.3.weight_compressed" type="Const" version="opset1">
243
+ <data element_type="f16" shape="64, 32, 3, 3" offset="1922" size="36864" />
244
+ <output>
245
+ <port id="0" precision="FP16" names="FeatureExtraction.ConvNet.3.weight">
246
+ <dim>64</dim>
247
+ <dim>32</dim>
248
+ <dim>3</dim>
249
+ <dim>3</dim>
250
+ </port>
251
+ </output>
252
+ </layer>
253
+ <layer id="15" name="FeatureExtraction.ConvNet.3.weight" type="Convert" version="opset1">
254
+ <data destination_type="f32" />
255
+ <rt_info>
256
+ <attribute name="decompression" version="0" />
257
+ </rt_info>
258
+ <input>
259
+ <port id="0" precision="FP16">
260
+ <dim>64</dim>
261
+ <dim>32</dim>
262
+ <dim>3</dim>
263
+ <dim>3</dim>
264
+ </port>
265
+ </input>
266
+ <output>
267
+ <port id="1" precision="FP32">
268
+ <dim>64</dim>
269
+ <dim>32</dim>
270
+ <dim>3</dim>
271
+ <dim>3</dim>
272
+ </port>
273
+ </output>
274
+ </layer>
275
+ <layer id="16" name="/FeatureExtraction/ConvNet/ConvNet.3/Conv/WithoutBiases" type="Convolution" version="opset1">
276
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
277
+ <input>
278
+ <port id="0" precision="FP32">
279
+ <dim>1</dim>
280
+ <dim>32</dim>
281
+ <dim>32</dim>
282
+ <dim>192</dim>
283
+ </port>
284
+ <port id="1" precision="FP32">
285
+ <dim>64</dim>
286
+ <dim>32</dim>
287
+ <dim>3</dim>
288
+ <dim>3</dim>
289
+ </port>
290
+ </input>
291
+ <output>
292
+ <port id="2" precision="FP32">
293
+ <dim>1</dim>
294
+ <dim>64</dim>
295
+ <dim>32</dim>
296
+ <dim>192</dim>
297
+ </port>
298
+ </output>
299
+ </layer>
300
+ <layer id="17" name="Reshape_273166_compressed" type="Const" version="opset1">
301
+ <data element_type="f16" shape="1, 64, 1, 1" offset="38786" size="128" />
302
+ <output>
303
+ <port id="0" precision="FP16">
304
+ <dim>1</dim>
305
+ <dim>64</dim>
306
+ <dim>1</dim>
307
+ <dim>1</dim>
308
+ </port>
309
+ </output>
310
+ </layer>
311
+ <layer id="18" name="Reshape_273166" type="Convert" version="opset1">
312
+ <data destination_type="f32" />
313
+ <rt_info>
314
+ <attribute name="decompression" version="0" />
315
+ </rt_info>
316
+ <input>
317
+ <port id="0" precision="FP16">
318
+ <dim>1</dim>
319
+ <dim>64</dim>
320
+ <dim>1</dim>
321
+ <dim>1</dim>
322
+ </port>
323
+ </input>
324
+ <output>
325
+ <port id="1" precision="FP32">
326
+ <dim>1</dim>
327
+ <dim>64</dim>
328
+ <dim>1</dim>
329
+ <dim>1</dim>
330
+ </port>
331
+ </output>
332
+ </layer>
333
+ <layer id="19" name="/FeatureExtraction/ConvNet/ConvNet.3/Conv" type="Add" version="opset1">
334
+ <data auto_broadcast="numpy" />
335
+ <input>
336
+ <port id="0" precision="FP32">
337
+ <dim>1</dim>
338
+ <dim>64</dim>
339
+ <dim>32</dim>
340
+ <dim>192</dim>
341
+ </port>
342
+ <port id="1" precision="FP32">
343
+ <dim>1</dim>
344
+ <dim>64</dim>
345
+ <dim>1</dim>
346
+ <dim>1</dim>
347
+ </port>
348
+ </input>
349
+ <output>
350
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.3/Conv_output_0">
351
+ <dim>1</dim>
352
+ <dim>64</dim>
353
+ <dim>32</dim>
354
+ <dim>192</dim>
355
+ </port>
356
+ </output>
357
+ </layer>
358
+ <layer id="20" name="/FeatureExtraction/ConvNet/ConvNet.4/Relu" type="ReLU" version="opset1">
359
+ <input>
360
+ <port id="0" precision="FP32">
361
+ <dim>1</dim>
362
+ <dim>64</dim>
363
+ <dim>32</dim>
364
+ <dim>192</dim>
365
+ </port>
366
+ </input>
367
+ <output>
368
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.4/Relu_output_0">
369
+ <dim>1</dim>
370
+ <dim>64</dim>
371
+ <dim>32</dim>
372
+ <dim>192</dim>
373
+ </port>
374
+ </output>
375
+ </layer>
376
+ <layer id="21" name="/FeatureExtraction/ConvNet/ConvNet.5/MaxPool" type="MaxPool" version="opset8">
377
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" kernel="2, 2" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
378
+ <input>
379
+ <port id="0" precision="FP32">
380
+ <dim>1</dim>
381
+ <dim>64</dim>
382
+ <dim>32</dim>
383
+ <dim>192</dim>
384
+ </port>
385
+ </input>
386
+ <output>
387
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.5/MaxPool_output_0">
388
+ <dim>1</dim>
389
+ <dim>64</dim>
390
+ <dim>16</dim>
391
+ <dim>96</dim>
392
+ </port>
393
+ <port id="2" precision="I64">
394
+ <dim>1</dim>
395
+ <dim>64</dim>
396
+ <dim>16</dim>
397
+ <dim>96</dim>
398
+ </port>
399
+ </output>
400
+ </layer>
401
+ <layer id="22" name="FeatureExtraction.ConvNet.6.weight_compressed" type="Const" version="opset1">
402
+ <data element_type="f16" shape="128, 64, 3, 3" offset="38914" size="147456" />
403
+ <output>
404
+ <port id="0" precision="FP16" names="FeatureExtraction.ConvNet.6.weight">
405
+ <dim>128</dim>
406
+ <dim>64</dim>
407
+ <dim>3</dim>
408
+ <dim>3</dim>
409
+ </port>
410
+ </output>
411
+ </layer>
412
+ <layer id="23" name="FeatureExtraction.ConvNet.6.weight" type="Convert" version="opset1">
413
+ <data destination_type="f32" />
414
+ <rt_info>
415
+ <attribute name="decompression" version="0" />
416
+ </rt_info>
417
+ <input>
418
+ <port id="0" precision="FP16">
419
+ <dim>128</dim>
420
+ <dim>64</dim>
421
+ <dim>3</dim>
422
+ <dim>3</dim>
423
+ </port>
424
+ </input>
425
+ <output>
426
+ <port id="1" precision="FP32">
427
+ <dim>128</dim>
428
+ <dim>64</dim>
429
+ <dim>3</dim>
430
+ <dim>3</dim>
431
+ </port>
432
+ </output>
433
+ </layer>
434
+ <layer id="24" name="/FeatureExtraction/ConvNet/ConvNet.6/Conv/WithoutBiases" type="Convolution" version="opset1">
435
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
436
+ <input>
437
+ <port id="0" precision="FP32">
438
+ <dim>1</dim>
439
+ <dim>64</dim>
440
+ <dim>16</dim>
441
+ <dim>96</dim>
442
+ </port>
443
+ <port id="1" precision="FP32">
444
+ <dim>128</dim>
445
+ <dim>64</dim>
446
+ <dim>3</dim>
447
+ <dim>3</dim>
448
+ </port>
449
+ </input>
450
+ <output>
451
+ <port id="2" precision="FP32">
452
+ <dim>1</dim>
453
+ <dim>128</dim>
454
+ <dim>16</dim>
455
+ <dim>96</dim>
456
+ </port>
457
+ </output>
458
+ </layer>
459
+ <layer id="25" name="Reshape_273183_compressed" type="Const" version="opset1">
460
+ <data element_type="f16" shape="1, 128, 1, 1" offset="186370" size="256" />
461
+ <output>
462
+ <port id="0" precision="FP16">
463
+ <dim>1</dim>
464
+ <dim>128</dim>
465
+ <dim>1</dim>
466
+ <dim>1</dim>
467
+ </port>
468
+ </output>
469
+ </layer>
470
+ <layer id="26" name="Reshape_273183" type="Convert" version="opset1">
471
+ <data destination_type="f32" />
472
+ <rt_info>
473
+ <attribute name="decompression" version="0" />
474
+ </rt_info>
475
+ <input>
476
+ <port id="0" precision="FP16">
477
+ <dim>1</dim>
478
+ <dim>128</dim>
479
+ <dim>1</dim>
480
+ <dim>1</dim>
481
+ </port>
482
+ </input>
483
+ <output>
484
+ <port id="1" precision="FP32">
485
+ <dim>1</dim>
486
+ <dim>128</dim>
487
+ <dim>1</dim>
488
+ <dim>1</dim>
489
+ </port>
490
+ </output>
491
+ </layer>
492
+ <layer id="27" name="/FeatureExtraction/ConvNet/ConvNet.6/Conv" type="Add" version="opset1">
493
+ <data auto_broadcast="numpy" />
494
+ <input>
495
+ <port id="0" precision="FP32">
496
+ <dim>1</dim>
497
+ <dim>128</dim>
498
+ <dim>16</dim>
499
+ <dim>96</dim>
500
+ </port>
501
+ <port id="1" precision="FP32">
502
+ <dim>1</dim>
503
+ <dim>128</dim>
504
+ <dim>1</dim>
505
+ <dim>1</dim>
506
+ </port>
507
+ </input>
508
+ <output>
509
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.6/Conv_output_0">
510
+ <dim>1</dim>
511
+ <dim>128</dim>
512
+ <dim>16</dim>
513
+ <dim>96</dim>
514
+ </port>
515
+ </output>
516
+ </layer>
517
+ <layer id="28" name="/FeatureExtraction/ConvNet/ConvNet.7/Relu" type="ReLU" version="opset1">
518
+ <input>
519
+ <port id="0" precision="FP32">
520
+ <dim>1</dim>
521
+ <dim>128</dim>
522
+ <dim>16</dim>
523
+ <dim>96</dim>
524
+ </port>
525
+ </input>
526
+ <output>
527
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.7/Relu_output_0">
528
+ <dim>1</dim>
529
+ <dim>128</dim>
530
+ <dim>16</dim>
531
+ <dim>96</dim>
532
+ </port>
533
+ </output>
534
+ </layer>
535
+ <layer id="29" name="FeatureExtraction.ConvNet.8.weight_compressed" type="Const" version="opset1">
536
+ <data element_type="f16" shape="128, 128, 3, 3" offset="186626" size="294912" />
537
+ <output>
538
+ <port id="0" precision="FP16" names="FeatureExtraction.ConvNet.8.weight">
539
+ <dim>128</dim>
540
+ <dim>128</dim>
541
+ <dim>3</dim>
542
+ <dim>3</dim>
543
+ </port>
544
+ </output>
545
+ </layer>
546
+ <layer id="30" name="FeatureExtraction.ConvNet.8.weight" type="Convert" version="opset1">
547
+ <data destination_type="f32" />
548
+ <rt_info>
549
+ <attribute name="decompression" version="0" />
550
+ </rt_info>
551
+ <input>
552
+ <port id="0" precision="FP16">
553
+ <dim>128</dim>
554
+ <dim>128</dim>
555
+ <dim>3</dim>
556
+ <dim>3</dim>
557
+ </port>
558
+ </input>
559
+ <output>
560
+ <port id="1" precision="FP32">
561
+ <dim>128</dim>
562
+ <dim>128</dim>
563
+ <dim>3</dim>
564
+ <dim>3</dim>
565
+ </port>
566
+ </output>
567
+ </layer>
568
+ <layer id="31" name="/FeatureExtraction/ConvNet/ConvNet.8/Conv/WithoutBiases" type="Convolution" version="opset1">
569
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
570
+ <input>
571
+ <port id="0" precision="FP32">
572
+ <dim>1</dim>
573
+ <dim>128</dim>
574
+ <dim>16</dim>
575
+ <dim>96</dim>
576
+ </port>
577
+ <port id="1" precision="FP32">
578
+ <dim>128</dim>
579
+ <dim>128</dim>
580
+ <dim>3</dim>
581
+ <dim>3</dim>
582
+ </port>
583
+ </input>
584
+ <output>
585
+ <port id="2" precision="FP32">
586
+ <dim>1</dim>
587
+ <dim>128</dim>
588
+ <dim>16</dim>
589
+ <dim>96</dim>
590
+ </port>
591
+ </output>
592
+ </layer>
593
+ <layer id="32" name="Reshape_273199" type="Const" version="opset1">
594
+ <data element_type="f32" shape="1, 128, 1, 1" offset="481538" size="512" />
595
+ <output>
596
+ <port id="0" precision="FP32">
597
+ <dim>1</dim>
598
+ <dim>128</dim>
599
+ <dim>1</dim>
600
+ <dim>1</dim>
601
+ </port>
602
+ </output>
603
+ </layer>
604
+ <layer id="33" name="/FeatureExtraction/ConvNet/ConvNet.8/Conv" type="Add" version="opset1">
605
+ <data auto_broadcast="numpy" />
606
+ <input>
607
+ <port id="0" precision="FP32">
608
+ <dim>1</dim>
609
+ <dim>128</dim>
610
+ <dim>16</dim>
611
+ <dim>96</dim>
612
+ </port>
613
+ <port id="1" precision="FP32">
614
+ <dim>1</dim>
615
+ <dim>128</dim>
616
+ <dim>1</dim>
617
+ <dim>1</dim>
618
+ </port>
619
+ </input>
620
+ <output>
621
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.8/Conv_output_0">
622
+ <dim>1</dim>
623
+ <dim>128</dim>
624
+ <dim>16</dim>
625
+ <dim>96</dim>
626
+ </port>
627
+ </output>
628
+ </layer>
629
+ <layer id="34" name="/FeatureExtraction/ConvNet/ConvNet.9/Relu" type="ReLU" version="opset1">
630
+ <input>
631
+ <port id="0" precision="FP32">
632
+ <dim>1</dim>
633
+ <dim>128</dim>
634
+ <dim>16</dim>
635
+ <dim>96</dim>
636
+ </port>
637
+ </input>
638
+ <output>
639
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.9/Relu_output_0">
640
+ <dim>1</dim>
641
+ <dim>128</dim>
642
+ <dim>16</dim>
643
+ <dim>96</dim>
644
+ </port>
645
+ </output>
646
+ </layer>
647
+ <layer id="35" name="/FeatureExtraction/ConvNet/ConvNet.10/MaxPool" type="MaxPool" version="opset8">
648
+ <data strides="2, 1" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" kernel="2, 1" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
649
+ <input>
650
+ <port id="0" precision="FP32">
651
+ <dim>1</dim>
652
+ <dim>128</dim>
653
+ <dim>16</dim>
654
+ <dim>96</dim>
655
+ </port>
656
+ </input>
657
+ <output>
658
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.10/MaxPool_output_0">
659
+ <dim>1</dim>
660
+ <dim>128</dim>
661
+ <dim>8</dim>
662
+ <dim>96</dim>
663
+ </port>
664
+ <port id="2" precision="I64">
665
+ <dim>1</dim>
666
+ <dim>128</dim>
667
+ <dim>8</dim>
668
+ <dim>96</dim>
669
+ </port>
670
+ </output>
671
+ </layer>
672
+ <layer id="36" name="onnx::Conv_385_compressed" type="Const" version="opset1">
673
+ <data element_type="f16" shape="256, 128, 3, 3" offset="482050" size="589824" />
674
+ <output>
675
+ <port id="0" precision="FP16" names="onnx::Conv_385">
676
+ <dim>256</dim>
677
+ <dim>128</dim>
678
+ <dim>3</dim>
679
+ <dim>3</dim>
680
+ </port>
681
+ </output>
682
+ </layer>
683
+ <layer id="37" name="onnx::Conv_385" type="Convert" version="opset1">
684
+ <data destination_type="f32" />
685
+ <rt_info>
686
+ <attribute name="decompression" version="0" />
687
+ </rt_info>
688
+ <input>
689
+ <port id="0" precision="FP16">
690
+ <dim>256</dim>
691
+ <dim>128</dim>
692
+ <dim>3</dim>
693
+ <dim>3</dim>
694
+ </port>
695
+ </input>
696
+ <output>
697
+ <port id="1" precision="FP32">
698
+ <dim>256</dim>
699
+ <dim>128</dim>
700
+ <dim>3</dim>
701
+ <dim>3</dim>
702
+ </port>
703
+ </output>
704
+ </layer>
705
+ <layer id="38" name="/FeatureExtraction/ConvNet/ConvNet.11/Conv/WithoutBiases" type="Convolution" version="opset1">
706
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
707
+ <input>
708
+ <port id="0" precision="FP32">
709
+ <dim>1</dim>
710
+ <dim>128</dim>
711
+ <dim>8</dim>
712
+ <dim>96</dim>
713
+ </port>
714
+ <port id="1" precision="FP32">
715
+ <dim>256</dim>
716
+ <dim>128</dim>
717
+ <dim>3</dim>
718
+ <dim>3</dim>
719
+ </port>
720
+ </input>
721
+ <output>
722
+ <port id="2" precision="FP32">
723
+ <dim>1</dim>
724
+ <dim>256</dim>
725
+ <dim>8</dim>
726
+ <dim>96</dim>
727
+ </port>
728
+ </output>
729
+ </layer>
730
+ <layer id="39" name="Reshape_273216_compressed" type="Const" version="opset1">
731
+ <data element_type="f16" shape="1, 256, 1, 1" offset="1071874" size="512" />
732
+ <output>
733
+ <port id="0" precision="FP16">
734
+ <dim>1</dim>
735
+ <dim>256</dim>
736
+ <dim>1</dim>
737
+ <dim>1</dim>
738
+ </port>
739
+ </output>
740
+ </layer>
741
+ <layer id="40" name="Reshape_273216" type="Convert" version="opset1">
742
+ <data destination_type="f32" />
743
+ <rt_info>
744
+ <attribute name="decompression" version="0" />
745
+ </rt_info>
746
+ <input>
747
+ <port id="0" precision="FP16">
748
+ <dim>1</dim>
749
+ <dim>256</dim>
750
+ <dim>1</dim>
751
+ <dim>1</dim>
752
+ </port>
753
+ </input>
754
+ <output>
755
+ <port id="1" precision="FP32">
756
+ <dim>1</dim>
757
+ <dim>256</dim>
758
+ <dim>1</dim>
759
+ <dim>1</dim>
760
+ </port>
761
+ </output>
762
+ </layer>
763
+ <layer id="41" name="/FeatureExtraction/ConvNet/ConvNet.11/Conv" type="Add" version="opset1">
764
+ <data auto_broadcast="numpy" />
765
+ <input>
766
+ <port id="0" precision="FP32">
767
+ <dim>1</dim>
768
+ <dim>256</dim>
769
+ <dim>8</dim>
770
+ <dim>96</dim>
771
+ </port>
772
+ <port id="1" precision="FP32">
773
+ <dim>1</dim>
774
+ <dim>256</dim>
775
+ <dim>1</dim>
776
+ <dim>1</dim>
777
+ </port>
778
+ </input>
779
+ <output>
780
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.11/Conv_output_0">
781
+ <dim>1</dim>
782
+ <dim>256</dim>
783
+ <dim>8</dim>
784
+ <dim>96</dim>
785
+ </port>
786
+ </output>
787
+ </layer>
788
+ <layer id="42" name="/FeatureExtraction/ConvNet/ConvNet.13/Relu" type="ReLU" version="opset1">
789
+ <input>
790
+ <port id="0" precision="FP32">
791
+ <dim>1</dim>
792
+ <dim>256</dim>
793
+ <dim>8</dim>
794
+ <dim>96</dim>
795
+ </port>
796
+ </input>
797
+ <output>
798
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.13/Relu_output_0">
799
+ <dim>1</dim>
800
+ <dim>256</dim>
801
+ <dim>8</dim>
802
+ <dim>96</dim>
803
+ </port>
804
+ </output>
805
+ </layer>
806
+ <layer id="43" name="onnx::Conv_388_compressed" type="Const" version="opset1">
807
+ <data element_type="f16" shape="256, 256, 3, 3" offset="1072386" size="1179648" />
808
+ <output>
809
+ <port id="0" precision="FP16" names="onnx::Conv_388">
810
+ <dim>256</dim>
811
+ <dim>256</dim>
812
+ <dim>3</dim>
813
+ <dim>3</dim>
814
+ </port>
815
+ </output>
816
+ </layer>
817
+ <layer id="44" name="onnx::Conv_388" type="Convert" version="opset1">
818
+ <data destination_type="f32" />
819
+ <rt_info>
820
+ <attribute name="decompression" version="0" />
821
+ </rt_info>
822
+ <input>
823
+ <port id="0" precision="FP16">
824
+ <dim>256</dim>
825
+ <dim>256</dim>
826
+ <dim>3</dim>
827
+ <dim>3</dim>
828
+ </port>
829
+ </input>
830
+ <output>
831
+ <port id="1" precision="FP32">
832
+ <dim>256</dim>
833
+ <dim>256</dim>
834
+ <dim>3</dim>
835
+ <dim>3</dim>
836
+ </port>
837
+ </output>
838
+ </layer>
839
+ <layer id="45" name="/FeatureExtraction/ConvNet/ConvNet.14/Conv/WithoutBiases" type="Convolution" version="opset1">
840
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
841
+ <input>
842
+ <port id="0" precision="FP32">
843
+ <dim>1</dim>
844
+ <dim>256</dim>
845
+ <dim>8</dim>
846
+ <dim>96</dim>
847
+ </port>
848
+ <port id="1" precision="FP32">
849
+ <dim>256</dim>
850
+ <dim>256</dim>
851
+ <dim>3</dim>
852
+ <dim>3</dim>
853
+ </port>
854
+ </input>
855
+ <output>
856
+ <port id="2" precision="FP32">
857
+ <dim>1</dim>
858
+ <dim>256</dim>
859
+ <dim>8</dim>
860
+ <dim>96</dim>
861
+ </port>
862
+ </output>
863
+ </layer>
864
+ <layer id="46" name="Reshape_273232_compressed" type="Const" version="opset1">
865
+ <data element_type="f16" shape="1, 256, 1, 1" offset="2252034" size="512" />
866
+ <output>
867
+ <port id="0" precision="FP16">
868
+ <dim>1</dim>
869
+ <dim>256</dim>
870
+ <dim>1</dim>
871
+ <dim>1</dim>
872
+ </port>
873
+ </output>
874
+ </layer>
875
+ <layer id="47" name="Reshape_273232" type="Convert" version="opset1">
876
+ <data destination_type="f32" />
877
+ <rt_info>
878
+ <attribute name="decompression" version="0" />
879
+ </rt_info>
880
+ <input>
881
+ <port id="0" precision="FP16">
882
+ <dim>1</dim>
883
+ <dim>256</dim>
884
+ <dim>1</dim>
885
+ <dim>1</dim>
886
+ </port>
887
+ </input>
888
+ <output>
889
+ <port id="1" precision="FP32">
890
+ <dim>1</dim>
891
+ <dim>256</dim>
892
+ <dim>1</dim>
893
+ <dim>1</dim>
894
+ </port>
895
+ </output>
896
+ </layer>
897
+ <layer id="48" name="/FeatureExtraction/ConvNet/ConvNet.14/Conv" type="Add" version="opset1">
898
+ <data auto_broadcast="numpy" />
899
+ <input>
900
+ <port id="0" precision="FP32">
901
+ <dim>1</dim>
902
+ <dim>256</dim>
903
+ <dim>8</dim>
904
+ <dim>96</dim>
905
+ </port>
906
+ <port id="1" precision="FP32">
907
+ <dim>1</dim>
908
+ <dim>256</dim>
909
+ <dim>1</dim>
910
+ <dim>1</dim>
911
+ </port>
912
+ </input>
913
+ <output>
914
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.14/Conv_output_0">
915
+ <dim>1</dim>
916
+ <dim>256</dim>
917
+ <dim>8</dim>
918
+ <dim>96</dim>
919
+ </port>
920
+ </output>
921
+ </layer>
922
+ <layer id="49" name="/FeatureExtraction/ConvNet/ConvNet.16/Relu" type="ReLU" version="opset1">
923
+ <input>
924
+ <port id="0" precision="FP32">
925
+ <dim>1</dim>
926
+ <dim>256</dim>
927
+ <dim>8</dim>
928
+ <dim>96</dim>
929
+ </port>
930
+ </input>
931
+ <output>
932
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.16/Relu_output_0">
933
+ <dim>1</dim>
934
+ <dim>256</dim>
935
+ <dim>8</dim>
936
+ <dim>96</dim>
937
+ </port>
938
+ </output>
939
+ </layer>
940
+ <layer id="50" name="/FeatureExtraction/ConvNet/ConvNet.17/MaxPool" type="MaxPool" version="opset8">
941
+ <data strides="2, 1" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" kernel="2, 1" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
942
+ <input>
943
+ <port id="0" precision="FP32">
944
+ <dim>1</dim>
945
+ <dim>256</dim>
946
+ <dim>8</dim>
947
+ <dim>96</dim>
948
+ </port>
949
+ </input>
950
+ <output>
951
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.17/MaxPool_output_0">
952
+ <dim>1</dim>
953
+ <dim>256</dim>
954
+ <dim>4</dim>
955
+ <dim>96</dim>
956
+ </port>
957
+ <port id="2" precision="I64">
958
+ <dim>1</dim>
959
+ <dim>256</dim>
960
+ <dim>4</dim>
961
+ <dim>96</dim>
962
+ </port>
963
+ </output>
964
+ </layer>
965
+ <layer id="51" name="FeatureExtraction.ConvNet.18.weight_compressed" type="Const" version="opset1">
966
+ <data element_type="f16" shape="256, 256, 2, 2" offset="2252546" size="524288" />
967
+ <output>
968
+ <port id="0" precision="FP16" names="FeatureExtraction.ConvNet.18.weight">
969
+ <dim>256</dim>
970
+ <dim>256</dim>
971
+ <dim>2</dim>
972
+ <dim>2</dim>
973
+ </port>
974
+ </output>
975
+ </layer>
976
+ <layer id="52" name="FeatureExtraction.ConvNet.18.weight" type="Convert" version="opset1">
977
+ <data destination_type="f32" />
978
+ <rt_info>
979
+ <attribute name="decompression" version="0" />
980
+ </rt_info>
981
+ <input>
982
+ <port id="0" precision="FP16">
983
+ <dim>256</dim>
984
+ <dim>256</dim>
985
+ <dim>2</dim>
986
+ <dim>2</dim>
987
+ </port>
988
+ </input>
989
+ <output>
990
+ <port id="1" precision="FP32">
991
+ <dim>256</dim>
992
+ <dim>256</dim>
993
+ <dim>2</dim>
994
+ <dim>2</dim>
995
+ </port>
996
+ </output>
997
+ </layer>
998
+ <layer id="53" name="/FeatureExtraction/ConvNet/ConvNet.18/Conv/WithoutBiases" type="Convolution" version="opset1">
999
+ <data strides="1, 1" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
1000
+ <input>
1001
+ <port id="0" precision="FP32">
1002
+ <dim>1</dim>
1003
+ <dim>256</dim>
1004
+ <dim>4</dim>
1005
+ <dim>96</dim>
1006
+ </port>
1007
+ <port id="1" precision="FP32">
1008
+ <dim>256</dim>
1009
+ <dim>256</dim>
1010
+ <dim>2</dim>
1011
+ <dim>2</dim>
1012
+ </port>
1013
+ </input>
1014
+ <output>
1015
+ <port id="2" precision="FP32">
1016
+ <dim>1</dim>
1017
+ <dim>256</dim>
1018
+ <dim>3</dim>
1019
+ <dim>95</dim>
1020
+ </port>
1021
+ </output>
1022
+ </layer>
1023
+ <layer id="54" name="Reshape_273249_compressed" type="Const" version="opset1">
1024
+ <data element_type="f16" shape="1, 256, 1, 1" offset="2776834" size="512" />
1025
+ <output>
1026
+ <port id="0" precision="FP16">
1027
+ <dim>1</dim>
1028
+ <dim>256</dim>
1029
+ <dim>1</dim>
1030
+ <dim>1</dim>
1031
+ </port>
1032
+ </output>
1033
+ </layer>
1034
+ <layer id="55" name="Reshape_273249" type="Convert" version="opset1">
1035
+ <data destination_type="f32" />
1036
+ <rt_info>
1037
+ <attribute name="decompression" version="0" />
1038
+ </rt_info>
1039
+ <input>
1040
+ <port id="0" precision="FP16">
1041
+ <dim>1</dim>
1042
+ <dim>256</dim>
1043
+ <dim>1</dim>
1044
+ <dim>1</dim>
1045
+ </port>
1046
+ </input>
1047
+ <output>
1048
+ <port id="1" precision="FP32">
1049
+ <dim>1</dim>
1050
+ <dim>256</dim>
1051
+ <dim>1</dim>
1052
+ <dim>1</dim>
1053
+ </port>
1054
+ </output>
1055
+ </layer>
1056
+ <layer id="56" name="/FeatureExtraction/ConvNet/ConvNet.18/Conv" type="Add" version="opset1">
1057
+ <data auto_broadcast="numpy" />
1058
+ <input>
1059
+ <port id="0" precision="FP32">
1060
+ <dim>1</dim>
1061
+ <dim>256</dim>
1062
+ <dim>3</dim>
1063
+ <dim>95</dim>
1064
+ </port>
1065
+ <port id="1" precision="FP32">
1066
+ <dim>1</dim>
1067
+ <dim>256</dim>
1068
+ <dim>1</dim>
1069
+ <dim>1</dim>
1070
+ </port>
1071
+ </input>
1072
+ <output>
1073
+ <port id="2" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.18/Conv_output_0">
1074
+ <dim>1</dim>
1075
+ <dim>256</dim>
1076
+ <dim>3</dim>
1077
+ <dim>95</dim>
1078
+ </port>
1079
+ </output>
1080
+ </layer>
1081
+ <layer id="57" name="/FeatureExtraction/ConvNet/ConvNet.19/Relu" type="ReLU" version="opset1">
1082
+ <input>
1083
+ <port id="0" precision="FP32">
1084
+ <dim>1</dim>
1085
+ <dim>256</dim>
1086
+ <dim>3</dim>
1087
+ <dim>95</dim>
1088
+ </port>
1089
+ </input>
1090
+ <output>
1091
+ <port id="1" precision="FP32" names="/FeatureExtraction/ConvNet/ConvNet.19/Relu_output_0">
1092
+ <dim>1</dim>
1093
+ <dim>256</dim>
1094
+ <dim>3</dim>
1095
+ <dim>95</dim>
1096
+ </port>
1097
+ </output>
1098
+ </layer>
1099
+ <layer id="58" name="Constant_273252" type="Const" version="opset1">
1100
+ <data element_type="i64" shape="4" offset="2777346" size="32" />
1101
+ <output>
1102
+ <port id="0" precision="I64">
1103
+ <dim>4</dim>
1104
+ </port>
1105
+ </output>
1106
+ </layer>
1107
+ <layer id="59" name="/Transpose" type="Transpose" version="opset1">
1108
+ <input>
1109
+ <port id="0" precision="FP32">
1110
+ <dim>1</dim>
1111
+ <dim>256</dim>
1112
+ <dim>3</dim>
1113
+ <dim>95</dim>
1114
+ </port>
1115
+ <port id="1" precision="I64">
1116
+ <dim>4</dim>
1117
+ </port>
1118
+ </input>
1119
+ <output>
1120
+ <port id="2" precision="FP32" names="/Transpose_output_0">
1121
+ <dim>1</dim>
1122
+ <dim>95</dim>
1123
+ <dim>256</dim>
1124
+ <dim>3</dim>
1125
+ </port>
1126
+ </output>
1127
+ </layer>
1128
+ <layer id="60" name="/AdaptiveAvgPool/AveragePool" type="AvgPool" version="opset1">
1129
+ <data kernel="1, 3" strides="1, 3" pads_begin="0, 0" pads_end="0, 0" exclude-pad="true" auto_pad="explicit" rounding_type="floor" />
1130
+ <input>
1131
+ <port id="0" precision="FP32">
1132
+ <dim>1</dim>
1133
+ <dim>95</dim>
1134
+ <dim>256</dim>
1135
+ <dim>3</dim>
1136
+ </port>
1137
+ </input>
1138
+ <output>
1139
+ <port id="1" precision="FP32" names="/AdaptiveAvgPool/AveragePool_output_0">
1140
+ <dim>1</dim>
1141
+ <dim>95</dim>
1142
+ <dim>256</dim>
1143
+ <dim>1</dim>
1144
+ </port>
1145
+ </output>
1146
+ </layer>
1147
+ <layer id="61" name="/Constant" type="Const" version="opset1">
1148
+ <data element_type="i64" shape="1" offset="2777378" size="8" />
1149
+ <output>
1150
+ <port id="0" precision="I64" names="/Constant_output_0">
1151
+ <dim>1</dim>
1152
+ </port>
1153
+ </output>
1154
+ </layer>
1155
+ <layer id="62" name="/Squeeze" type="Squeeze" version="opset1">
1156
+ <input>
1157
+ <port id="0" precision="FP32">
1158
+ <dim>1</dim>
1159
+ <dim>95</dim>
1160
+ <dim>256</dim>
1161
+ <dim>1</dim>
1162
+ </port>
1163
+ <port id="1" precision="I64">
1164
+ <dim>1</dim>
1165
+ </port>
1166
+ </input>
1167
+ <output>
1168
+ <port id="2" precision="FP32" names="/Squeeze_output_0">
1169
+ <dim>1</dim>
1170
+ <dim>95</dim>
1171
+ <dim>256</dim>
1172
+ </port>
1173
+ </output>
1174
+ </layer>
1175
+ <layer id="63" name="Constant_273256" type="Const" version="opset1">
1176
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1177
+ <output>
1178
+ <port id="0" precision="I64">
1179
+ <dim>3</dim>
1180
+ </port>
1181
+ </output>
1182
+ </layer>
1183
+ <layer id="64" name="/SequenceModeling/SequenceModeling.0/rnn/Transpose" type="Transpose" version="opset1">
1184
+ <input>
1185
+ <port id="0" precision="FP32">
1186
+ <dim>1</dim>
1187
+ <dim>95</dim>
1188
+ <dim>256</dim>
1189
+ </port>
1190
+ <port id="1" precision="I64">
1191
+ <dim>3</dim>
1192
+ </port>
1193
+ </input>
1194
+ <output>
1195
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/rnn/Transpose_output_0">
1196
+ <dim>95</dim>
1197
+ <dim>1</dim>
1198
+ <dim>256</dim>
1199
+ </port>
1200
+ </output>
1201
+ </layer>
1202
+ <layer id="65" name="Constant_273278" type="Const" version="opset1">
1203
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1204
+ <output>
1205
+ <port id="0" precision="I64">
1206
+ <dim>3</dim>
1207
+ </port>
1208
+ </output>
1209
+ </layer>
1210
+ <layer id="66" name="Transpose_273279" type="Transpose" version="opset1">
1211
+ <input>
1212
+ <port id="0" precision="FP32">
1213
+ <dim>95</dim>
1214
+ <dim>1</dim>
1215
+ <dim>256</dim>
1216
+ </port>
1217
+ <port id="1" precision="I64">
1218
+ <dim>3</dim>
1219
+ </port>
1220
+ </input>
1221
+ <output>
1222
+ <port id="2" precision="FP32">
1223
+ <dim>1</dim>
1224
+ <dim>95</dim>
1225
+ <dim>256</dim>
1226
+ </port>
1227
+ </output>
1228
+ </layer>
1229
+ <layer id="67" name="/SequenceModeling/SequenceModeling.0/rnn/Constant_compressed" type="Const" version="opset1">
1230
+ <data element_type="f16" shape="2, 1, 256" offset="2777410" size="1024" />
1231
+ <output>
1232
+ <port id="0" precision="FP16" names="/SequenceModeling/SequenceModeling.0/rnn/Constant_output_0">
1233
+ <dim>2</dim>
1234
+ <dim>1</dim>
1235
+ <dim>256</dim>
1236
+ </port>
1237
+ </output>
1238
+ </layer>
1239
+ <layer id="68" name="/SequenceModeling/SequenceModeling.0/rnn/Constant" type="Convert" version="opset1">
1240
+ <data destination_type="f32" />
1241
+ <rt_info>
1242
+ <attribute name="decompression" version="0" />
1243
+ </rt_info>
1244
+ <input>
1245
+ <port id="0" precision="FP16">
1246
+ <dim>2</dim>
1247
+ <dim>1</dim>
1248
+ <dim>256</dim>
1249
+ </port>
1250
+ </input>
1251
+ <output>
1252
+ <port id="1" precision="FP32">
1253
+ <dim>2</dim>
1254
+ <dim>1</dim>
1255
+ <dim>256</dim>
1256
+ </port>
1257
+ </output>
1258
+ </layer>
1259
+ <layer id="69" name="Constant_51" type="Const" version="opset1">
1260
+ <data element_type="i64" shape="1" offset="2778434" size="8" />
1261
+ <rt_info>
1262
+ <attribute name="precise" version="0" />
1263
+ </rt_info>
1264
+ <output>
1265
+ <port id="0" precision="I64" names="onnx::Concat_433">
1266
+ <dim>1</dim>
1267
+ </port>
1268
+ </output>
1269
+ </layer>
1270
+ <layer id="70" name="ShapeOf_275754" type="ShapeOf" version="opset3">
1271
+ <data output_type="i64" />
1272
+ <input>
1273
+ <port id="0" precision="FP32">
1274
+ <dim>95</dim>
1275
+ <dim>1</dim>
1276
+ <dim>256</dim>
1277
+ </port>
1278
+ </input>
1279
+ <output>
1280
+ <port id="1" precision="I64">
1281
+ <dim>3</dim>
1282
+ </port>
1283
+ </output>
1284
+ </layer>
1285
+ <layer id="71" name="Constant_275755" type="Const" version="opset1">
1286
+ <data element_type="i64" shape="1" offset="2778442" size="8" />
1287
+ <rt_info>
1288
+ <attribute name="precise" version="0" />
1289
+ </rt_info>
1290
+ <output>
1291
+ <port id="0" precision="I64">
1292
+ <dim>1</dim>
1293
+ </port>
1294
+ </output>
1295
+ </layer>
1296
+ <layer id="72" name="Constant_275756" type="Const" version="opset1">
1297
+ <data element_type="i64" shape="" offset="2778450" size="8" />
1298
+ <rt_info>
1299
+ <attribute name="precise" version="0" />
1300
+ </rt_info>
1301
+ <output>
1302
+ <port id="0" precision="I64" />
1303
+ </output>
1304
+ </layer>
1305
+ <layer id="73" name="Gather_275757" type="Gather" version="opset8">
1306
+ <data batch_dims="0" />
1307
+ <input>
1308
+ <port id="0" precision="I64">
1309
+ <dim>3</dim>
1310
+ </port>
1311
+ <port id="1" precision="I64">
1312
+ <dim>1</dim>
1313
+ </port>
1314
+ <port id="2" precision="I64" />
1315
+ </input>
1316
+ <output>
1317
+ <port id="3" precision="I64" names="/SequenceModeling/SequenceModeling.0/rnn/Gather_1_output_0,/SequenceModeling/SequenceModeling.0/rnn/Gather_output_0,/SequenceModeling/SequenceModeling.1/rnn/Gather_1_output_0,/SequenceModeling/SequenceModeling.1/rnn/Gather_output_0,onnx::Concat_197,onnx::Concat_208,onnx::Concat_353,onnx::Concat_364">
1318
+ <dim>1</dim>
1319
+ </port>
1320
+ </output>
1321
+ </layer>
1322
+ <layer id="74" name="/SequenceModeling/SequenceModeling.0/rnn/Constant_2" type="Const" version="opset1">
1323
+ <data element_type="i64" shape="1" offset="2778458" size="8" />
1324
+ <rt_info>
1325
+ <attribute name="precise" version="0" />
1326
+ </rt_info>
1327
+ <output>
1328
+ <port id="0" precision="I64" names="/SequenceModeling/SequenceModeling.0/rnn/Constant_2_output_0">
1329
+ <dim>1</dim>
1330
+ </port>
1331
+ </output>
1332
+ </layer>
1333
+ <layer id="75" name="/SequenceModeling/SequenceModeling.0/rnn/Concat" type="Concat" version="opset1">
1334
+ <data axis="0" />
1335
+ <input>
1336
+ <port id="0" precision="I64">
1337
+ <dim>1</dim>
1338
+ </port>
1339
+ <port id="1" precision="I64">
1340
+ <dim>1</dim>
1341
+ </port>
1342
+ <port id="2" precision="I64">
1343
+ <dim>1</dim>
1344
+ </port>
1345
+ </input>
1346
+ <output>
1347
+ <port id="3" precision="I64" names="/SequenceModeling/SequenceModeling.0/rnn/Concat_1_output_0,/SequenceModeling/SequenceModeling.0/rnn/Concat_output_0,/SequenceModeling/SequenceModeling.1/rnn/Concat_1_output_0,/SequenceModeling/SequenceModeling.1/rnn/Concat_output_0">
1348
+ <dim>3</dim>
1349
+ </port>
1350
+ </output>
1351
+ </layer>
1352
+ <layer id="76" name="/SequenceModeling/SequenceModeling.0/rnn/Expand" type="Broadcast" version="opset3">
1353
+ <data mode="bidirectional" />
1354
+ <input>
1355
+ <port id="0" precision="FP32">
1356
+ <dim>2</dim>
1357
+ <dim>1</dim>
1358
+ <dim>256</dim>
1359
+ </port>
1360
+ <port id="1" precision="I64">
1361
+ <dim>3</dim>
1362
+ </port>
1363
+ </input>
1364
+ <output>
1365
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/rnn/Expand_1_output_0,/SequenceModeling/SequenceModeling.0/rnn/Expand_output_0,/SequenceModeling/SequenceModeling.1/rnn/Expand_1_output_0,/SequenceModeling/SequenceModeling.1/rnn/Expand_output_0">
1366
+ <dim>2</dim>
1367
+ <dim>1</dim>
1368
+ <dim>256</dim>
1369
+ </port>
1370
+ </output>
1371
+ </layer>
1372
+ <layer id="77" name="Constant_273306" type="Const" version="opset1">
1373
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1374
+ <output>
1375
+ <port id="0" precision="I64">
1376
+ <dim>3</dim>
1377
+ </port>
1378
+ </output>
1379
+ </layer>
1380
+ <layer id="78" name="Transpose_273307" type="Transpose" version="opset1">
1381
+ <input>
1382
+ <port id="0" precision="FP32">
1383
+ <dim>2</dim>
1384
+ <dim>1</dim>
1385
+ <dim>256</dim>
1386
+ </port>
1387
+ <port id="1" precision="I64">
1388
+ <dim>3</dim>
1389
+ </port>
1390
+ </input>
1391
+ <output>
1392
+ <port id="2" precision="FP32">
1393
+ <dim>1</dim>
1394
+ <dim>2</dim>
1395
+ <dim>256</dim>
1396
+ </port>
1397
+ </output>
1398
+ </layer>
1399
+ <layer id="79" name="ShapeOf_273286" type="ShapeOf" version="opset3">
1400
+ <data output_type="i64" />
1401
+ <input>
1402
+ <port id="0" precision="FP32">
1403
+ <dim>1</dim>
1404
+ <dim>95</dim>
1405
+ <dim>256</dim>
1406
+ </port>
1407
+ </input>
1408
+ <output>
1409
+ <port id="1" precision="I64">
1410
+ <dim>3</dim>
1411
+ </port>
1412
+ </output>
1413
+ </layer>
1414
+ <layer id="80" name="Constant_273290" type="Const" version="opset1">
1415
+ <data element_type="i32" shape="1" offset="2778466" size="4" />
1416
+ <output>
1417
+ <port id="0" precision="I32">
1418
+ <dim>1</dim>
1419
+ </port>
1420
+ </output>
1421
+ </layer>
1422
+ <layer id="81" name="Constant_273287" type="Const" version="opset1">
1423
+ <data element_type="i32" shape="1" offset="2778470" size="4" />
1424
+ <output>
1425
+ <port id="0" precision="I32">
1426
+ <dim>1</dim>
1427
+ </port>
1428
+ </output>
1429
+ </layer>
1430
+ <layer id="82" name="Gather_273291" type="Gather" version="opset8">
1431
+ <data batch_dims="0" />
1432
+ <input>
1433
+ <port id="0" precision="I64">
1434
+ <dim>3</dim>
1435
+ </port>
1436
+ <port id="1" precision="I32">
1437
+ <dim>1</dim>
1438
+ </port>
1439
+ <port id="2" precision="I32">
1440
+ <dim>1</dim>
1441
+ </port>
1442
+ </input>
1443
+ <output>
1444
+ <port id="3" precision="I64">
1445
+ <dim>1</dim>
1446
+ </port>
1447
+ </output>
1448
+ </layer>
1449
+ <layer id="83" name="Broadcast_273303" type="Broadcast" version="opset3">
1450
+ <data mode="numpy" />
1451
+ <input>
1452
+ <port id="0" precision="I64">
1453
+ <dim>1</dim>
1454
+ </port>
1455
+ <port id="1" precision="I64">
1456
+ <dim>1</dim>
1457
+ </port>
1458
+ </input>
1459
+ <output>
1460
+ <port id="2" precision="I64">
1461
+ <dim>1</dim>
1462
+ </port>
1463
+ </output>
1464
+ </layer>
1465
+ <layer id="84" name="Concat_273282_compressed" type="Const" version="opset1">
1466
+ <data element_type="f16" shape="2, 1024, 256" offset="2778474" size="1048576" />
1467
+ <output>
1468
+ <port id="0" precision="FP16">
1469
+ <dim>2</dim>
1470
+ <dim>1024</dim>
1471
+ <dim>256</dim>
1472
+ </port>
1473
+ </output>
1474
+ </layer>
1475
+ <layer id="85" name="Concat_273282" type="Convert" version="opset1">
1476
+ <data destination_type="f32" />
1477
+ <rt_info>
1478
+ <attribute name="decompression" version="0" />
1479
+ </rt_info>
1480
+ <input>
1481
+ <port id="0" precision="FP16">
1482
+ <dim>2</dim>
1483
+ <dim>1024</dim>
1484
+ <dim>256</dim>
1485
+ </port>
1486
+ </input>
1487
+ <output>
1488
+ <port id="1" precision="FP32">
1489
+ <dim>2</dim>
1490
+ <dim>1024</dim>
1491
+ <dim>256</dim>
1492
+ </port>
1493
+ </output>
1494
+ </layer>
1495
+ <layer id="86" name="Concat_273285_compressed" type="Const" version="opset1">
1496
+ <data element_type="f16" shape="2, 1024, 256" offset="3827050" size="1048576" />
1497
+ <output>
1498
+ <port id="0" precision="FP16">
1499
+ <dim>2</dim>
1500
+ <dim>1024</dim>
1501
+ <dim>256</dim>
1502
+ </port>
1503
+ </output>
1504
+ </layer>
1505
+ <layer id="87" name="Concat_273285" type="Convert" version="opset1">
1506
+ <data destination_type="f32" />
1507
+ <rt_info>
1508
+ <attribute name="decompression" version="0" />
1509
+ </rt_info>
1510
+ <input>
1511
+ <port id="0" precision="FP16">
1512
+ <dim>2</dim>
1513
+ <dim>1024</dim>
1514
+ <dim>256</dim>
1515
+ </port>
1516
+ </input>
1517
+ <output>
1518
+ <port id="1" precision="FP32">
1519
+ <dim>2</dim>
1520
+ <dim>1024</dim>
1521
+ <dim>256</dim>
1522
+ </port>
1523
+ </output>
1524
+ </layer>
1525
+ <layer id="88" name="Concat_273302_compressed" type="Const" version="opset1">
1526
+ <data element_type="f16" shape="2, 1024" offset="4875626" size="4096" />
1527
+ <output>
1528
+ <port id="0" precision="FP16">
1529
+ <dim>2</dim>
1530
+ <dim>1024</dim>
1531
+ </port>
1532
+ </output>
1533
+ </layer>
1534
+ <layer id="89" name="Concat_273302" type="Convert" version="opset1">
1535
+ <data destination_type="f32" />
1536
+ <rt_info>
1537
+ <attribute name="decompression" version="0" />
1538
+ </rt_info>
1539
+ <input>
1540
+ <port id="0" precision="FP16">
1541
+ <dim>2</dim>
1542
+ <dim>1024</dim>
1543
+ </port>
1544
+ </input>
1545
+ <output>
1546
+ <port id="1" precision="FP32">
1547
+ <dim>2</dim>
1548
+ <dim>1024</dim>
1549
+ </port>
1550
+ </output>
1551
+ </layer>
1552
+ <layer id="90" name="LSTMSequence_273317" type="LSTMSequence" version="opset5">
1553
+ <data direction="bidirectional" hidden_size="256" activations="sigmoid, tanh, tanh" activations_alpha="" activations_beta="" clip="0" />
1554
+ <input>
1555
+ <port id="0" precision="FP32">
1556
+ <dim>1</dim>
1557
+ <dim>95</dim>
1558
+ <dim>256</dim>
1559
+ </port>
1560
+ <port id="1" precision="FP32">
1561
+ <dim>1</dim>
1562
+ <dim>2</dim>
1563
+ <dim>256</dim>
1564
+ </port>
1565
+ <port id="2" precision="FP32">
1566
+ <dim>1</dim>
1567
+ <dim>2</dim>
1568
+ <dim>256</dim>
1569
+ </port>
1570
+ <port id="3" precision="I64">
1571
+ <dim>1</dim>
1572
+ </port>
1573
+ <port id="4" precision="FP32">
1574
+ <dim>2</dim>
1575
+ <dim>1024</dim>
1576
+ <dim>256</dim>
1577
+ </port>
1578
+ <port id="5" precision="FP32">
1579
+ <dim>2</dim>
1580
+ <dim>1024</dim>
1581
+ <dim>256</dim>
1582
+ </port>
1583
+ <port id="6" precision="FP32">
1584
+ <dim>2</dim>
1585
+ <dim>1024</dim>
1586
+ </port>
1587
+ </input>
1588
+ <output>
1589
+ <port id="7" precision="FP32">
1590
+ <dim>1</dim>
1591
+ <dim>2</dim>
1592
+ <dim>95</dim>
1593
+ <dim>256</dim>
1594
+ </port>
1595
+ <port id="8" precision="FP32">
1596
+ <dim>1</dim>
1597
+ <dim>2</dim>
1598
+ <dim>256</dim>
1599
+ </port>
1600
+ <port id="9" precision="FP32">
1601
+ <dim>1</dim>
1602
+ <dim>2</dim>
1603
+ <dim>256</dim>
1604
+ </port>
1605
+ </output>
1606
+ </layer>
1607
+ <layer id="91" name="Constant_275276" type="Const" version="opset1">
1608
+ <data element_type="i64" shape="4" offset="4879722" size="32" />
1609
+ <output>
1610
+ <port id="0" precision="I64">
1611
+ <dim>4</dim>
1612
+ </port>
1613
+ </output>
1614
+ </layer>
1615
+ <layer id="92" name="/SequenceModeling/SequenceModeling.0/rnn/Transpose_1" type="Transpose" version="opset1">
1616
+ <input>
1617
+ <port id="0" precision="FP32">
1618
+ <dim>1</dim>
1619
+ <dim>2</dim>
1620
+ <dim>95</dim>
1621
+ <dim>256</dim>
1622
+ </port>
1623
+ <port id="1" precision="I64">
1624
+ <dim>4</dim>
1625
+ </port>
1626
+ </input>
1627
+ <output>
1628
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/rnn/Transpose_1_output_0">
1629
+ <dim>95</dim>
1630
+ <dim>1</dim>
1631
+ <dim>2</dim>
1632
+ <dim>256</dim>
1633
+ </port>
1634
+ </output>
1635
+ </layer>
1636
+ <layer id="93" name="/SequenceModeling/SequenceModeling.0/rnn/Constant_5" type="Const" version="opset1">
1637
+ <data element_type="i64" shape="3" offset="4879754" size="24" />
1638
+ <rt_info>
1639
+ <attribute name="precise" version="0" />
1640
+ </rt_info>
1641
+ <output>
1642
+ <port id="0" precision="I64" names="/SequenceModeling/SequenceModeling.0/rnn/Constant_5_output_0">
1643
+ <dim>3</dim>
1644
+ </port>
1645
+ </output>
1646
+ </layer>
1647
+ <layer id="94" name="/SequenceModeling/SequenceModeling.0/rnn/Reshape" type="Reshape" version="opset1">
1648
+ <data special_zero="true" />
1649
+ <input>
1650
+ <port id="0" precision="FP32">
1651
+ <dim>95</dim>
1652
+ <dim>1</dim>
1653
+ <dim>2</dim>
1654
+ <dim>256</dim>
1655
+ </port>
1656
+ <port id="1" precision="I64">
1657
+ <dim>3</dim>
1658
+ </port>
1659
+ </input>
1660
+ <output>
1661
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/rnn/Reshape_output_0">
1662
+ <dim>95</dim>
1663
+ <dim>1</dim>
1664
+ <dim>512</dim>
1665
+ </port>
1666
+ </output>
1667
+ </layer>
1668
+ <layer id="95" name="Constant_273327" type="Const" version="opset1">
1669
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1670
+ <output>
1671
+ <port id="0" precision="I64">
1672
+ <dim>3</dim>
1673
+ </port>
1674
+ </output>
1675
+ </layer>
1676
+ <layer id="96" name="/SequenceModeling/SequenceModeling.0/rnn/Transpose_2" type="Transpose" version="opset1">
1677
+ <input>
1678
+ <port id="0" precision="FP32">
1679
+ <dim>95</dim>
1680
+ <dim>1</dim>
1681
+ <dim>512</dim>
1682
+ </port>
1683
+ <port id="1" precision="I64">
1684
+ <dim>3</dim>
1685
+ </port>
1686
+ </input>
1687
+ <output>
1688
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/rnn/Transpose_2_output_0">
1689
+ <dim>1</dim>
1690
+ <dim>95</dim>
1691
+ <dim>512</dim>
1692
+ </port>
1693
+ </output>
1694
+ </layer>
1695
+ <layer id="97" name="Transpose_275311_compressed" type="Const" version="opset1">
1696
+ <data element_type="f16" shape="256, 512" offset="4879778" size="262144" />
1697
+ <output>
1698
+ <port id="0" precision="FP16">
1699
+ <dim>256</dim>
1700
+ <dim>512</dim>
1701
+ </port>
1702
+ </output>
1703
+ </layer>
1704
+ <layer id="98" name="Transpose_275311" type="Convert" version="opset1">
1705
+ <data destination_type="f32" />
1706
+ <rt_info>
1707
+ <attribute name="decompression" version="0" />
1708
+ </rt_info>
1709
+ <input>
1710
+ <port id="0" precision="FP16">
1711
+ <dim>256</dim>
1712
+ <dim>512</dim>
1713
+ </port>
1714
+ </input>
1715
+ <output>
1716
+ <port id="1" precision="FP32">
1717
+ <dim>256</dim>
1718
+ <dim>512</dim>
1719
+ </port>
1720
+ </output>
1721
+ </layer>
1722
+ <layer id="99" name="/SequenceModeling/SequenceModeling.0/linear/MatMul" type="MatMul" version="opset1">
1723
+ <data transpose_a="false" transpose_b="true" />
1724
+ <input>
1725
+ <port id="0" precision="FP32">
1726
+ <dim>1</dim>
1727
+ <dim>95</dim>
1728
+ <dim>512</dim>
1729
+ </port>
1730
+ <port id="1" precision="FP32">
1731
+ <dim>256</dim>
1732
+ <dim>512</dim>
1733
+ </port>
1734
+ </input>
1735
+ <output>
1736
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/linear/MatMul_output_0">
1737
+ <dim>1</dim>
1738
+ <dim>95</dim>
1739
+ <dim>256</dim>
1740
+ </port>
1741
+ </output>
1742
+ </layer>
1743
+ <layer id="100" name="/SequenceModeling/SequenceModeling.0/linear/Add" type="Add" version="opset1">
1744
+ <data auto_broadcast="numpy" />
1745
+ <input>
1746
+ <port id="0" precision="FP32">
1747
+ <dim>1</dim>
1748
+ <dim>1</dim>
1749
+ <dim>256</dim>
1750
+ </port>
1751
+ <port id="1" precision="FP32">
1752
+ <dim>1</dim>
1753
+ <dim>95</dim>
1754
+ <dim>256</dim>
1755
+ </port>
1756
+ </input>
1757
+ <output>
1758
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.0/linear/Add_output_0">
1759
+ <dim>1</dim>
1760
+ <dim>95</dim>
1761
+ <dim>256</dim>
1762
+ </port>
1763
+ </output>
1764
+ </layer>
1765
+ <layer id="101" name="Constant_273333" type="Const" version="opset1">
1766
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1767
+ <output>
1768
+ <port id="0" precision="I64">
1769
+ <dim>3</dim>
1770
+ </port>
1771
+ </output>
1772
+ </layer>
1773
+ <layer id="102" name="/SequenceModeling/SequenceModeling.1/rnn/Transpose" type="Transpose" version="opset1">
1774
+ <input>
1775
+ <port id="0" precision="FP32">
1776
+ <dim>1</dim>
1777
+ <dim>95</dim>
1778
+ <dim>256</dim>
1779
+ </port>
1780
+ <port id="1" precision="I64">
1781
+ <dim>3</dim>
1782
+ </port>
1783
+ </input>
1784
+ <output>
1785
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/rnn/Transpose_output_0">
1786
+ <dim>95</dim>
1787
+ <dim>1</dim>
1788
+ <dim>256</dim>
1789
+ </port>
1790
+ </output>
1791
+ </layer>
1792
+ <layer id="103" name="Constant_273355" type="Const" version="opset1">
1793
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
1794
+ <output>
1795
+ <port id="0" precision="I64">
1796
+ <dim>3</dim>
1797
+ </port>
1798
+ </output>
1799
+ </layer>
1800
+ <layer id="104" name="Transpose_273356" type="Transpose" version="opset1">
1801
+ <input>
1802
+ <port id="0" precision="FP32">
1803
+ <dim>95</dim>
1804
+ <dim>1</dim>
1805
+ <dim>256</dim>
1806
+ </port>
1807
+ <port id="1" precision="I64">
1808
+ <dim>3</dim>
1809
+ </port>
1810
+ </input>
1811
+ <output>
1812
+ <port id="2" precision="FP32">
1813
+ <dim>1</dim>
1814
+ <dim>95</dim>
1815
+ <dim>256</dim>
1816
+ </port>
1817
+ </output>
1818
+ </layer>
1819
+ <layer id="105" name="Concat_273359_compressed" type="Const" version="opset1">
1820
+ <data element_type="f16" shape="2, 1024, 256" offset="5141922" size="1048576" />
1821
+ <output>
1822
+ <port id="0" precision="FP16">
1823
+ <dim>2</dim>
1824
+ <dim>1024</dim>
1825
+ <dim>256</dim>
1826
+ </port>
1827
+ </output>
1828
+ </layer>
1829
+ <layer id="106" name="Concat_273359" type="Convert" version="opset1">
1830
+ <data destination_type="f32" />
1831
+ <rt_info>
1832
+ <attribute name="decompression" version="0" />
1833
+ </rt_info>
1834
+ <input>
1835
+ <port id="0" precision="FP16">
1836
+ <dim>2</dim>
1837
+ <dim>1024</dim>
1838
+ <dim>256</dim>
1839
+ </port>
1840
+ </input>
1841
+ <output>
1842
+ <port id="1" precision="FP32">
1843
+ <dim>2</dim>
1844
+ <dim>1024</dim>
1845
+ <dim>256</dim>
1846
+ </port>
1847
+ </output>
1848
+ </layer>
1849
+ <layer id="107" name="Concat_273362_compressed" type="Const" version="opset1">
1850
+ <data element_type="f16" shape="2, 1024, 256" offset="6190498" size="1048576" />
1851
+ <output>
1852
+ <port id="0" precision="FP16">
1853
+ <dim>2</dim>
1854
+ <dim>1024</dim>
1855
+ <dim>256</dim>
1856
+ </port>
1857
+ </output>
1858
+ </layer>
1859
+ <layer id="108" name="Concat_273362" type="Convert" version="opset1">
1860
+ <data destination_type="f32" />
1861
+ <rt_info>
1862
+ <attribute name="decompression" version="0" />
1863
+ </rt_info>
1864
+ <input>
1865
+ <port id="0" precision="FP16">
1866
+ <dim>2</dim>
1867
+ <dim>1024</dim>
1868
+ <dim>256</dim>
1869
+ </port>
1870
+ </input>
1871
+ <output>
1872
+ <port id="1" precision="FP32">
1873
+ <dim>2</dim>
1874
+ <dim>1024</dim>
1875
+ <dim>256</dim>
1876
+ </port>
1877
+ </output>
1878
+ </layer>
1879
+ <layer id="109" name="Concat_273379_compressed" type="Const" version="opset1">
1880
+ <data element_type="f16" shape="2, 1024" offset="7239074" size="4096" />
1881
+ <output>
1882
+ <port id="0" precision="FP16">
1883
+ <dim>2</dim>
1884
+ <dim>1024</dim>
1885
+ </port>
1886
+ </output>
1887
+ </layer>
1888
+ <layer id="110" name="Concat_273379" type="Convert" version="opset1">
1889
+ <data destination_type="f32" />
1890
+ <rt_info>
1891
+ <attribute name="decompression" version="0" />
1892
+ </rt_info>
1893
+ <input>
1894
+ <port id="0" precision="FP16">
1895
+ <dim>2</dim>
1896
+ <dim>1024</dim>
1897
+ </port>
1898
+ </input>
1899
+ <output>
1900
+ <port id="1" precision="FP32">
1901
+ <dim>2</dim>
1902
+ <dim>1024</dim>
1903
+ </port>
1904
+ </output>
1905
+ </layer>
1906
+ <layer id="111" name="LSTMSequence_273394" type="LSTMSequence" version="opset5">
1907
+ <data direction="bidirectional" hidden_size="256" activations="sigmoid, tanh, tanh" activations_alpha="" activations_beta="" clip="0" />
1908
+ <input>
1909
+ <port id="0" precision="FP32">
1910
+ <dim>1</dim>
1911
+ <dim>95</dim>
1912
+ <dim>256</dim>
1913
+ </port>
1914
+ <port id="1" precision="FP32">
1915
+ <dim>1</dim>
1916
+ <dim>2</dim>
1917
+ <dim>256</dim>
1918
+ </port>
1919
+ <port id="2" precision="FP32">
1920
+ <dim>1</dim>
1921
+ <dim>2</dim>
1922
+ <dim>256</dim>
1923
+ </port>
1924
+ <port id="3" precision="I64">
1925
+ <dim>1</dim>
1926
+ </port>
1927
+ <port id="4" precision="FP32">
1928
+ <dim>2</dim>
1929
+ <dim>1024</dim>
1930
+ <dim>256</dim>
1931
+ </port>
1932
+ <port id="5" precision="FP32">
1933
+ <dim>2</dim>
1934
+ <dim>1024</dim>
1935
+ <dim>256</dim>
1936
+ </port>
1937
+ <port id="6" precision="FP32">
1938
+ <dim>2</dim>
1939
+ <dim>1024</dim>
1940
+ </port>
1941
+ </input>
1942
+ <output>
1943
+ <port id="7" precision="FP32">
1944
+ <dim>1</dim>
1945
+ <dim>2</dim>
1946
+ <dim>95</dim>
1947
+ <dim>256</dim>
1948
+ </port>
1949
+ <port id="8" precision="FP32">
1950
+ <dim>1</dim>
1951
+ <dim>2</dim>
1952
+ <dim>256</dim>
1953
+ </port>
1954
+ <port id="9" precision="FP32">
1955
+ <dim>1</dim>
1956
+ <dim>2</dim>
1957
+ <dim>256</dim>
1958
+ </port>
1959
+ </output>
1960
+ </layer>
1961
+ <layer id="112" name="Constant_275278" type="Const" version="opset1">
1962
+ <data element_type="i64" shape="4" offset="4879722" size="32" />
1963
+ <output>
1964
+ <port id="0" precision="I64">
1965
+ <dim>4</dim>
1966
+ </port>
1967
+ </output>
1968
+ </layer>
1969
+ <layer id="113" name="/SequenceModeling/SequenceModeling.1/rnn/Transpose_1" type="Transpose" version="opset1">
1970
+ <input>
1971
+ <port id="0" precision="FP32">
1972
+ <dim>1</dim>
1973
+ <dim>2</dim>
1974
+ <dim>95</dim>
1975
+ <dim>256</dim>
1976
+ </port>
1977
+ <port id="1" precision="I64">
1978
+ <dim>4</dim>
1979
+ </port>
1980
+ </input>
1981
+ <output>
1982
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/rnn/Transpose_1_output_0">
1983
+ <dim>95</dim>
1984
+ <dim>1</dim>
1985
+ <dim>2</dim>
1986
+ <dim>256</dim>
1987
+ </port>
1988
+ </output>
1989
+ </layer>
1990
+ <layer id="114" name="/SequenceModeling/SequenceModeling.1/rnn/Constant_5" type="Const" version="opset1">
1991
+ <data element_type="i64" shape="3" offset="4879754" size="24" />
1992
+ <rt_info>
1993
+ <attribute name="precise" version="0" />
1994
+ </rt_info>
1995
+ <output>
1996
+ <port id="0" precision="I64" names="/SequenceModeling/SequenceModeling.1/rnn/Constant_5_output_0">
1997
+ <dim>3</dim>
1998
+ </port>
1999
+ </output>
2000
+ </layer>
2001
+ <layer id="115" name="/SequenceModeling/SequenceModeling.1/rnn/Reshape" type="Reshape" version="opset1">
2002
+ <data special_zero="true" />
2003
+ <input>
2004
+ <port id="0" precision="FP32">
2005
+ <dim>95</dim>
2006
+ <dim>1</dim>
2007
+ <dim>2</dim>
2008
+ <dim>256</dim>
2009
+ </port>
2010
+ <port id="1" precision="I64">
2011
+ <dim>3</dim>
2012
+ </port>
2013
+ </input>
2014
+ <output>
2015
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/rnn/Reshape_output_0">
2016
+ <dim>95</dim>
2017
+ <dim>1</dim>
2018
+ <dim>512</dim>
2019
+ </port>
2020
+ </output>
2021
+ </layer>
2022
+ <layer id="116" name="Constant_273404" type="Const" version="opset1">
2023
+ <data element_type="i64" shape="3" offset="2777386" size="24" />
2024
+ <output>
2025
+ <port id="0" precision="I64">
2026
+ <dim>3</dim>
2027
+ </port>
2028
+ </output>
2029
+ </layer>
2030
+ <layer id="117" name="/SequenceModeling/SequenceModeling.1/rnn/Transpose_2" type="Transpose" version="opset1">
2031
+ <input>
2032
+ <port id="0" precision="FP32">
2033
+ <dim>95</dim>
2034
+ <dim>1</dim>
2035
+ <dim>512</dim>
2036
+ </port>
2037
+ <port id="1" precision="I64">
2038
+ <dim>3</dim>
2039
+ </port>
2040
+ </input>
2041
+ <output>
2042
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/rnn/Transpose_2_output_0">
2043
+ <dim>1</dim>
2044
+ <dim>95</dim>
2045
+ <dim>512</dim>
2046
+ </port>
2047
+ </output>
2048
+ </layer>
2049
+ <layer id="118" name="Transpose_275318_compressed" type="Const" version="opset1">
2050
+ <data element_type="f16" shape="256, 512" offset="7243170" size="262144" />
2051
+ <output>
2052
+ <port id="0" precision="FP16">
2053
+ <dim>256</dim>
2054
+ <dim>512</dim>
2055
+ </port>
2056
+ </output>
2057
+ </layer>
2058
+ <layer id="119" name="Transpose_275318" type="Convert" version="opset1">
2059
+ <data destination_type="f32" />
2060
+ <rt_info>
2061
+ <attribute name="decompression" version="0" />
2062
+ </rt_info>
2063
+ <input>
2064
+ <port id="0" precision="FP16">
2065
+ <dim>256</dim>
2066
+ <dim>512</dim>
2067
+ </port>
2068
+ </input>
2069
+ <output>
2070
+ <port id="1" precision="FP32">
2071
+ <dim>256</dim>
2072
+ <dim>512</dim>
2073
+ </port>
2074
+ </output>
2075
+ </layer>
2076
+ <layer id="120" name="/SequenceModeling/SequenceModeling.1/linear/MatMul" type="MatMul" version="opset1">
2077
+ <data transpose_a="false" transpose_b="true" />
2078
+ <input>
2079
+ <port id="0" precision="FP32">
2080
+ <dim>1</dim>
2081
+ <dim>95</dim>
2082
+ <dim>512</dim>
2083
+ </port>
2084
+ <port id="1" precision="FP32">
2085
+ <dim>256</dim>
2086
+ <dim>512</dim>
2087
+ </port>
2088
+ </input>
2089
+ <output>
2090
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/linear/MatMul_output_0">
2091
+ <dim>1</dim>
2092
+ <dim>95</dim>
2093
+ <dim>256</dim>
2094
+ </port>
2095
+ </output>
2096
+ </layer>
2097
+ <layer id="121" name="/SequenceModeling/SequenceModeling.1/linear/Add" type="Add" version="opset1">
2098
+ <data auto_broadcast="numpy" />
2099
+ <input>
2100
+ <port id="0" precision="FP32">
2101
+ <dim>1</dim>
2102
+ <dim>1</dim>
2103
+ <dim>256</dim>
2104
+ </port>
2105
+ <port id="1" precision="FP32">
2106
+ <dim>1</dim>
2107
+ <dim>95</dim>
2108
+ <dim>256</dim>
2109
+ </port>
2110
+ </input>
2111
+ <output>
2112
+ <port id="2" precision="FP32" names="/SequenceModeling/SequenceModeling.1/linear/Add_output_0">
2113
+ <dim>1</dim>
2114
+ <dim>95</dim>
2115
+ <dim>256</dim>
2116
+ </port>
2117
+ </output>
2118
+ </layer>
2119
+ <layer id="122" name="Transpose_275325_compressed" type="Const" version="opset1">
2120
+ <data element_type="f16" shape="97, 256" offset="7505314" size="49664" />
2121
+ <output>
2122
+ <port id="0" precision="FP16">
2123
+ <dim>97</dim>
2124
+ <dim>256</dim>
2125
+ </port>
2126
+ </output>
2127
+ </layer>
2128
+ <layer id="123" name="Transpose_275325" type="Convert" version="opset1">
2129
+ <data destination_type="f32" />
2130
+ <rt_info>
2131
+ <attribute name="decompression" version="0" />
2132
+ </rt_info>
2133
+ <input>
2134
+ <port id="0" precision="FP16">
2135
+ <dim>97</dim>
2136
+ <dim>256</dim>
2137
+ </port>
2138
+ </input>
2139
+ <output>
2140
+ <port id="1" precision="FP32">
2141
+ <dim>97</dim>
2142
+ <dim>256</dim>
2143
+ </port>
2144
+ </output>
2145
+ </layer>
2146
+ <layer id="124" name="/Prediction/MatMul" type="MatMul" version="opset1">
2147
+ <data transpose_a="false" transpose_b="true" />
2148
+ <input>
2149
+ <port id="0" precision="FP32">
2150
+ <dim>1</dim>
2151
+ <dim>95</dim>
2152
+ <dim>256</dim>
2153
+ </port>
2154
+ <port id="1" precision="FP32">
2155
+ <dim>97</dim>
2156
+ <dim>256</dim>
2157
+ </port>
2158
+ </input>
2159
+ <output>
2160
+ <port id="2" precision="FP32" names="/Prediction/MatMul_output_0">
2161
+ <dim>1</dim>
2162
+ <dim>95</dim>
2163
+ <dim>97</dim>
2164
+ </port>
2165
+ </output>
2166
+ </layer>
2167
+ <layer id="125" name="383" type="Add" version="opset1">
2168
+ <data auto_broadcast="numpy" />
2169
+ <input>
2170
+ <port id="0" precision="FP32">
2171
+ <dim>1</dim>
2172
+ <dim>1</dim>
2173
+ <dim>97</dim>
2174
+ </port>
2175
+ <port id="1" precision="FP32">
2176
+ <dim>1</dim>
2177
+ <dim>95</dim>
2178
+ <dim>97</dim>
2179
+ </port>
2180
+ </input>
2181
+ <output>
2182
+ <port id="2" precision="FP32" names="383">
2183
+ <dim>1</dim>
2184
+ <dim>95</dim>
2185
+ <dim>97</dim>
2186
+ </port>
2187
+ </output>
2188
+ </layer>
2189
+ <layer id="126" name="383/sink_port_0" type="Result" version="opset1" output_names="383">
2190
+ <input>
2191
+ <port id="0" precision="FP32">
2192
+ <dim>1</dim>
2193
+ <dim>95</dim>
2194
+ <dim>97</dim>
2195
+ </port>
2196
+ </input>
2197
+ </layer>
2198
+ </layers>
2199
+ <edges>
2200
+ <edge from-layer="0" from-port="0" to-layer="9" to-port="0" />
2201
+ <edge from-layer="1" from-port="0" to-layer="2" to-port="0" />
2202
+ <edge from-layer="2" from-port="1" to-layer="125" to-port="0" />
2203
+ <edge from-layer="3" from-port="0" to-layer="4" to-port="0" />
2204
+ <edge from-layer="4" from-port="1" to-layer="121" to-port="0" />
2205
+ <edge from-layer="5" from-port="0" to-layer="6" to-port="0" />
2206
+ <edge from-layer="6" from-port="1" to-layer="100" to-port="0" />
2207
+ <edge from-layer="7" from-port="0" to-layer="8" to-port="0" />
2208
+ <edge from-layer="8" from-port="1" to-layer="9" to-port="1" />
2209
+ <edge from-layer="9" from-port="2" to-layer="11" to-port="0" />
2210
+ <edge from-layer="10" from-port="0" to-layer="11" to-port="1" />
2211
+ <edge from-layer="11" from-port="2" to-layer="12" to-port="0" />
2212
+ <edge from-layer="12" from-port="1" to-layer="13" to-port="0" />
2213
+ <edge from-layer="13" from-port="1" to-layer="16" to-port="0" />
2214
+ <edge from-layer="14" from-port="0" to-layer="15" to-port="0" />
2215
+ <edge from-layer="15" from-port="1" to-layer="16" to-port="1" />
2216
+ <edge from-layer="16" from-port="2" to-layer="19" to-port="0" />
2217
+ <edge from-layer="17" from-port="0" to-layer="18" to-port="0" />
2218
+ <edge from-layer="18" from-port="1" to-layer="19" to-port="1" />
2219
+ <edge from-layer="19" from-port="2" to-layer="20" to-port="0" />
2220
+ <edge from-layer="20" from-port="1" to-layer="21" to-port="0" />
2221
+ <edge from-layer="21" from-port="1" to-layer="24" to-port="0" />
2222
+ <edge from-layer="22" from-port="0" to-layer="23" to-port="0" />
2223
+ <edge from-layer="23" from-port="1" to-layer="24" to-port="1" />
2224
+ <edge from-layer="24" from-port="2" to-layer="27" to-port="0" />
2225
+ <edge from-layer="25" from-port="0" to-layer="26" to-port="0" />
2226
+ <edge from-layer="26" from-port="1" to-layer="27" to-port="1" />
2227
+ <edge from-layer="27" from-port="2" to-layer="28" to-port="0" />
2228
+ <edge from-layer="28" from-port="1" to-layer="31" to-port="0" />
2229
+ <edge from-layer="29" from-port="0" to-layer="30" to-port="0" />
2230
+ <edge from-layer="30" from-port="1" to-layer="31" to-port="1" />
2231
+ <edge from-layer="31" from-port="2" to-layer="33" to-port="0" />
2232
+ <edge from-layer="32" from-port="0" to-layer="33" to-port="1" />
2233
+ <edge from-layer="33" from-port="2" to-layer="34" to-port="0" />
2234
+ <edge from-layer="34" from-port="1" to-layer="35" to-port="0" />
2235
+ <edge from-layer="35" from-port="1" to-layer="38" to-port="0" />
2236
+ <edge from-layer="36" from-port="0" to-layer="37" to-port="0" />
2237
+ <edge from-layer="37" from-port="1" to-layer="38" to-port="1" />
2238
+ <edge from-layer="38" from-port="2" to-layer="41" to-port="0" />
2239
+ <edge from-layer="39" from-port="0" to-layer="40" to-port="0" />
2240
+ <edge from-layer="40" from-port="1" to-layer="41" to-port="1" />
2241
+ <edge from-layer="41" from-port="2" to-layer="42" to-port="0" />
2242
+ <edge from-layer="42" from-port="1" to-layer="45" to-port="0" />
2243
+ <edge from-layer="43" from-port="0" to-layer="44" to-port="0" />
2244
+ <edge from-layer="44" from-port="1" to-layer="45" to-port="1" />
2245
+ <edge from-layer="45" from-port="2" to-layer="48" to-port="0" />
2246
+ <edge from-layer="46" from-port="0" to-layer="47" to-port="0" />
2247
+ <edge from-layer="47" from-port="1" to-layer="48" to-port="1" />
2248
+ <edge from-layer="48" from-port="2" to-layer="49" to-port="0" />
2249
+ <edge from-layer="49" from-port="1" to-layer="50" to-port="0" />
2250
+ <edge from-layer="50" from-port="1" to-layer="53" to-port="0" />
2251
+ <edge from-layer="51" from-port="0" to-layer="52" to-port="0" />
2252
+ <edge from-layer="52" from-port="1" to-layer="53" to-port="1" />
2253
+ <edge from-layer="53" from-port="2" to-layer="56" to-port="0" />
2254
+ <edge from-layer="54" from-port="0" to-layer="55" to-port="0" />
2255
+ <edge from-layer="55" from-port="1" to-layer="56" to-port="1" />
2256
+ <edge from-layer="56" from-port="2" to-layer="57" to-port="0" />
2257
+ <edge from-layer="57" from-port="1" to-layer="59" to-port="0" />
2258
+ <edge from-layer="58" from-port="0" to-layer="59" to-port="1" />
2259
+ <edge from-layer="59" from-port="2" to-layer="60" to-port="0" />
2260
+ <edge from-layer="60" from-port="1" to-layer="62" to-port="0" />
2261
+ <edge from-layer="61" from-port="0" to-layer="62" to-port="1" />
2262
+ <edge from-layer="62" from-port="2" to-layer="64" to-port="0" />
2263
+ <edge from-layer="63" from-port="0" to-layer="64" to-port="1" />
2264
+ <edge from-layer="64" from-port="2" to-layer="66" to-port="0" />
2265
+ <edge from-layer="64" from-port="2" to-layer="70" to-port="0" />
2266
+ <edge from-layer="65" from-port="0" to-layer="66" to-port="1" />
2267
+ <edge from-layer="66" from-port="2" to-layer="90" to-port="0" />
2268
+ <edge from-layer="66" from-port="2" to-layer="79" to-port="0" />
2269
+ <edge from-layer="67" from-port="0" to-layer="68" to-port="0" />
2270
+ <edge from-layer="68" from-port="1" to-layer="76" to-port="0" />
2271
+ <edge from-layer="69" from-port="0" to-layer="75" to-port="0" />
2272
+ <edge from-layer="70" from-port="1" to-layer="73" to-port="0" />
2273
+ <edge from-layer="71" from-port="0" to-layer="73" to-port="1" />
2274
+ <edge from-layer="72" from-port="0" to-layer="73" to-port="2" />
2275
+ <edge from-layer="73" from-port="3" to-layer="75" to-port="1" />
2276
+ <edge from-layer="73" from-port="3" to-layer="83" to-port="1" />
2277
+ <edge from-layer="74" from-port="0" to-layer="75" to-port="2" />
2278
+ <edge from-layer="75" from-port="3" to-layer="76" to-port="1" />
2279
+ <edge from-layer="76" from-port="2" to-layer="78" to-port="0" />
2280
+ <edge from-layer="77" from-port="0" to-layer="78" to-port="1" />
2281
+ <edge from-layer="78" from-port="2" to-layer="111" to-port="1" />
2282
+ <edge from-layer="78" from-port="2" to-layer="90" to-port="1" />
2283
+ <edge from-layer="78" from-port="2" to-layer="90" to-port="2" />
2284
+ <edge from-layer="78" from-port="2" to-layer="111" to-port="2" />
2285
+ <edge from-layer="79" from-port="1" to-layer="82" to-port="0" />
2286
+ <edge from-layer="80" from-port="0" to-layer="82" to-port="1" />
2287
+ <edge from-layer="81" from-port="0" to-layer="82" to-port="2" />
2288
+ <edge from-layer="82" from-port="3" to-layer="83" to-port="0" />
2289
+ <edge from-layer="83" from-port="2" to-layer="90" to-port="3" />
2290
+ <edge from-layer="83" from-port="2" to-layer="111" to-port="3" />
2291
+ <edge from-layer="84" from-port="0" to-layer="85" to-port="0" />
2292
+ <edge from-layer="85" from-port="1" to-layer="90" to-port="4" />
2293
+ <edge from-layer="86" from-port="0" to-layer="87" to-port="0" />
2294
+ <edge from-layer="87" from-port="1" to-layer="90" to-port="5" />
2295
+ <edge from-layer="88" from-port="0" to-layer="89" to-port="0" />
2296
+ <edge from-layer="89" from-port="1" to-layer="90" to-port="6" />
2297
+ <edge from-layer="90" from-port="7" to-layer="92" to-port="0" />
2298
+ <edge from-layer="91" from-port="0" to-layer="92" to-port="1" />
2299
+ <edge from-layer="92" from-port="2" to-layer="94" to-port="0" />
2300
+ <edge from-layer="93" from-port="0" to-layer="94" to-port="1" />
2301
+ <edge from-layer="94" from-port="2" to-layer="96" to-port="0" />
2302
+ <edge from-layer="95" from-port="0" to-layer="96" to-port="1" />
2303
+ <edge from-layer="96" from-port="2" to-layer="99" to-port="0" />
2304
+ <edge from-layer="97" from-port="0" to-layer="98" to-port="0" />
2305
+ <edge from-layer="98" from-port="1" to-layer="99" to-port="1" />
2306
+ <edge from-layer="99" from-port="2" to-layer="100" to-port="1" />
2307
+ <edge from-layer="100" from-port="2" to-layer="102" to-port="0" />
2308
+ <edge from-layer="101" from-port="0" to-layer="102" to-port="1" />
2309
+ <edge from-layer="102" from-port="2" to-layer="104" to-port="0" />
2310
+ <edge from-layer="103" from-port="0" to-layer="104" to-port="1" />
2311
+ <edge from-layer="104" from-port="2" to-layer="111" to-port="0" />
2312
+ <edge from-layer="105" from-port="0" to-layer="106" to-port="0" />
2313
+ <edge from-layer="106" from-port="1" to-layer="111" to-port="4" />
2314
+ <edge from-layer="107" from-port="0" to-layer="108" to-port="0" />
2315
+ <edge from-layer="108" from-port="1" to-layer="111" to-port="5" />
2316
+ <edge from-layer="109" from-port="0" to-layer="110" to-port="0" />
2317
+ <edge from-layer="110" from-port="1" to-layer="111" to-port="6" />
2318
+ <edge from-layer="111" from-port="7" to-layer="113" to-port="0" />
2319
+ <edge from-layer="112" from-port="0" to-layer="113" to-port="1" />
2320
+ <edge from-layer="113" from-port="2" to-layer="115" to-port="0" />
2321
+ <edge from-layer="114" from-port="0" to-layer="115" to-port="1" />
2322
+ <edge from-layer="115" from-port="2" to-layer="117" to-port="0" />
2323
+ <edge from-layer="116" from-port="0" to-layer="117" to-port="1" />
2324
+ <edge from-layer="117" from-port="2" to-layer="120" to-port="0" />
2325
+ <edge from-layer="118" from-port="0" to-layer="119" to-port="0" />
2326
+ <edge from-layer="119" from-port="1" to-layer="120" to-port="1" />
2327
+ <edge from-layer="120" from-port="2" to-layer="121" to-port="1" />
2328
+ <edge from-layer="121" from-port="2" to-layer="124" to-port="0" />
2329
+ <edge from-layer="122" from-port="0" to-layer="123" to-port="0" />
2330
+ <edge from-layer="123" from-port="1" to-layer="124" to-port="1" />
2331
+ <edge from-layer="124" from-port="2" to-layer="125" to-port="1" />
2332
+ <edge from-layer="125" from-port="2" to-layer="126" to-port="0" />
2333
+ </edges>
2334
+ <rt_info>
2335
+ <info name="OpenVINO Runtime" value="2026.2.1-21919-ede283a88e3-releases/2026/2" />
2336
+ <Runtime_version value="2026.2.1-21919-ede283a88e3-releases/2026/2" />
2337
+ <conversion_parameters>
2338
+ <input_model value="DIR/vgg_english_g2.onnx" />
2339
+ <is_python_object value="False" />
2340
+ </conversion_parameters>
2341
+ </rt_info>
2342
+ </net>