Hugging Face's logo

camstack
/
camstack-models

TF-Keras
LiteRT
Core ML
ONNX
Model card Files
xet
 Community
1
apocaliss92 commited on
Commit
0810df6
·
verified ·
1 Parent(s): cb2e649

add animalClassification/animal-cls-v1 small classifier (onnx+openvino+coreml+labels)

Browse files
Files changed (10) hide show
  1. animalClassification/animal-cls-v1/animal-cls-v1-labels.json +35 -0
  2. animalClassification/animal-cls-v1/coreml/animal-cls-v1-labels.json +35 -0
  3. animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Data/com.apple.CoreML/model.mlmodel +3 -0
  4. animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Data/com.apple.CoreML/weights/weight.bin +3 -0
  5. animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Manifest.json +18 -0
  6. animalClassification/animal-cls-v1/onnx/animal-cls-v1-labels.json +35 -0
  7. animalClassification/animal-cls-v1/onnx/animal-cls-v1.onnx +3 -0
  8. animalClassification/animal-cls-v1/openvino/animal-cls-v1-labels.json +35 -0
  9. animalClassification/animal-cls-v1/openvino/animal-cls-v1.bin +3 -0
  10. animalClassification/animal-cls-v1/openvino/animal-cls-v1.xml +3188 -0
animalClassification/animal-cls-v1/animal-cls-v1-labels.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "butterfly",
4
+ "cat",
5
+ "chicken",
6
+ "cow",
7
+ "dog",
8
+ "elephant",
9
+ "horse",
10
+ "sheep",
11
+ "spider",
12
+ "squirrel"
13
+ ],
14
+ "input_size": 224,
15
+ "num_classes": 10,
16
+ "preprocess": {
17
+ "normalize": "imagenet",
18
+ "mean": [
19
+ 0.485,
20
+ 0.456,
21
+ 0.406
22
+ ],
23
+ "std": [
24
+ 0.229,
25
+ 0.224,
26
+ 0.225
27
+ ],
28
+ "channels": "RGB",
29
+ "layout": "NCHW"
30
+ },
31
+ "arch": "resnet18",
32
+ "source": "MichaelMM2000/animals10-resnet",
33
+ "license": "MIT",
34
+ "output": "logits (apply softmax)"
35
+ }
animalClassification/animal-cls-v1/coreml/animal-cls-v1-labels.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "butterfly",
4
+ "cat",
5
+ "chicken",
6
+ "cow",
7
+ "dog",
8
+ "elephant",
9
+ "horse",
10
+ "sheep",
11
+ "spider",
12
+ "squirrel"
13
+ ],
14
+ "input_size": 224,
15
+ "num_classes": 10,
16
+ "preprocess": {
17
+ "normalize": "imagenet",
18
+ "mean": [
19
+ 0.485,
20
+ 0.456,
21
+ 0.406
22
+ ],
23
+ "std": [
24
+ 0.229,
25
+ 0.224,
26
+ 0.225
27
+ ],
28
+ "channels": "RGB",
29
+ "layout": "NCHW"
30
+ },
31
+ "arch": "resnet18",
32
+ "source": "MichaelMM2000/animals10-resnet",
33
+ "license": "MIT",
34
+ "output": "logits (apply softmax)"
35
+ }
animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Data/com.apple.CoreML/model.mlmodel ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa3fb006f9b312817c18f103b28c3d1c0575f4bec20f3b3c1c618b666b5522a7
3
+ size 31463
animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Data/com.apple.CoreML/weights/weight.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35b29a2c5c2a2b740ffe0cd30e71414b7ef48f55aa872d66921dc5c1d57d6b26
3
+ size 22356436
animalClassification/animal-cls-v1/coreml/animal-cls-v1.mlpackage/Manifest.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "fileFormatVersion": "1.0.0",
3
+ "itemInfoEntries": {
4
+ "6F88DD0A-DF0C-4246-87DC-AAEA4213B54A": {
5
+ "author": "com.apple.CoreML",
6
+ "description": "CoreML Model Specification",
7
+ "name": "model.mlmodel",
8
+ "path": "com.apple.CoreML/model.mlmodel"
9
+ },
10
+ "A4D44BE7-19F6-44C8-A250-6108EE36B255": {
11
+ "author": "com.apple.CoreML",
12
+ "description": "CoreML Model Weights",
13
+ "name": "weights",
14
+ "path": "com.apple.CoreML/weights"
15
+ }
16
+ },
17
+ "rootModelIdentifier": "6F88DD0A-DF0C-4246-87DC-AAEA4213B54A"
18
+ }
animalClassification/animal-cls-v1/onnx/animal-cls-v1-labels.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "butterfly",
4
+ "cat",
5
+ "chicken",
6
+ "cow",
7
+ "dog",
8
+ "elephant",
9
+ "horse",
10
+ "sheep",
11
+ "spider",
12
+ "squirrel"
13
+ ],
14
+ "input_size": 224,
15
+ "num_classes": 10,
16
+ "preprocess": {
17
+ "normalize": "imagenet",
18
+ "mean": [
19
+ 0.485,
20
+ 0.456,
21
+ 0.406
22
+ ],
23
+ "std": [
24
+ 0.229,
25
+ 0.224,
26
+ 0.225
27
+ ],
28
+ "channels": "RGB",
29
+ "layout": "NCHW"
30
+ },
31
+ "arch": "resnet18",
32
+ "source": "MichaelMM2000/animals10-resnet",
33
+ "license": "MIT",
34
+ "output": "logits (apply softmax)"
35
+ }
animalClassification/animal-cls-v1/onnx/animal-cls-v1.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48731f9848c9077651e704df32a7677b17030341520bfe58349ccc07cca04278
3
+ size 44717227
animalClassification/animal-cls-v1/openvino/animal-cls-v1-labels.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "butterfly",
4
+ "cat",
5
+ "chicken",
6
+ "cow",
7
+ "dog",
8
+ "elephant",
9
+ "horse",
10
+ "sheep",
11
+ "spider",
12
+ "squirrel"
13
+ ],
14
+ "input_size": 224,
15
+ "num_classes": 10,
16
+ "preprocess": {
17
+ "normalize": "imagenet",
18
+ "mean": [
19
+ 0.485,
20
+ 0.456,
21
+ 0.406
22
+ ],
23
+ "std": [
24
+ 0.229,
25
+ 0.224,
26
+ 0.225
27
+ ],
28
+ "channels": "RGB",
29
+ "layout": "NCHW"
30
+ },
31
+ "arch": "resnet18",
32
+ "source": "MichaelMM2000/animals10-resnet",
33
+ "license": "MIT",
34
+ "output": "logits (apply softmax)"
35
+ }
animalClassification/animal-cls-v1/openvino/animal-cls-v1.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eddb683aeb9f2a3a893c75ba8267a938fe5d2a115e517d1b0920e89f96cb58fc
3
+ size 22353736
animalClassification/animal-cls-v1/openvino/animal-cls-v1.xml ADDED
@@ -0,0 +1,3188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <?xml version="1.0"?>
2
+ <net name="onnx_Frontend_IR" version="11">
3
+ <layers>
4
+ <layer id="0" name="input" type="Parameter" version="opset1">
5
+ <data shape="?,3,224,224" element_type="f32" />
6
+ <output>
7
+ <port id="0" precision="FP32" names="input">
8
+ <dim>-1</dim>
9
+ <dim>3</dim>
10
+ <dim>224</dim>
11
+ <dim>224</dim>
12
+ </port>
13
+ </output>
14
+ </layer>
15
+ <layer id="1" name="onnx::Conv_193_compressed" type="Const" version="opset1">
16
+ <data element_type="f16" shape="64, 3, 7, 7" offset="0" size="18816" />
17
+ <output>
18
+ <port id="0" precision="FP16" names="onnx::Conv_193">
19
+ <dim>64</dim>
20
+ <dim>3</dim>
21
+ <dim>7</dim>
22
+ <dim>7</dim>
23
+ </port>
24
+ </output>
25
+ </layer>
26
+ <layer id="2" name="onnx::Conv_193" type="Convert" version="opset1">
27
+ <data destination_type="f32" />
28
+ <rt_info>
29
+ <attribute name="decompression" version="0" />
30
+ </rt_info>
31
+ <input>
32
+ <port id="0" precision="FP16">
33
+ <dim>64</dim>
34
+ <dim>3</dim>
35
+ <dim>7</dim>
36
+ <dim>7</dim>
37
+ </port>
38
+ </input>
39
+ <output>
40
+ <port id="1" precision="FP32">
41
+ <dim>64</dim>
42
+ <dim>3</dim>
43
+ <dim>7</dim>
44
+ <dim>7</dim>
45
+ </port>
46
+ </output>
47
+ </layer>
48
+ <layer id="3" name="/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
49
+ <data strides="2, 2" dilations="1, 1" pads_begin="3, 3" pads_end="3, 3" auto_pad="explicit" />
50
+ <input>
51
+ <port id="0" precision="FP32">
52
+ <dim>-1</dim>
53
+ <dim>3</dim>
54
+ <dim>224</dim>
55
+ <dim>224</dim>
56
+ </port>
57
+ <port id="1" precision="FP32">
58
+ <dim>64</dim>
59
+ <dim>3</dim>
60
+ <dim>7</dim>
61
+ <dim>7</dim>
62
+ </port>
63
+ </input>
64
+ <output>
65
+ <port id="2" precision="FP32">
66
+ <dim>-1</dim>
67
+ <dim>64</dim>
68
+ <dim>112</dim>
69
+ <dim>112</dim>
70
+ </port>
71
+ </output>
72
+ </layer>
73
+ <layer id="4" name="Reshape_15_compressed" type="Const" version="opset1">
74
+ <data element_type="f16" shape="1, 64, 1, 1" offset="18816" size="128" />
75
+ <output>
76
+ <port id="0" precision="FP16">
77
+ <dim>1</dim>
78
+ <dim>64</dim>
79
+ <dim>1</dim>
80
+ <dim>1</dim>
81
+ </port>
82
+ </output>
83
+ </layer>
84
+ <layer id="5" name="Reshape_15" type="Convert" version="opset1">
85
+ <data destination_type="f32" />
86
+ <rt_info>
87
+ <attribute name="decompression" version="0" />
88
+ </rt_info>
89
+ <input>
90
+ <port id="0" precision="FP16">
91
+ <dim>1</dim>
92
+ <dim>64</dim>
93
+ <dim>1</dim>
94
+ <dim>1</dim>
95
+ </port>
96
+ </input>
97
+ <output>
98
+ <port id="1" precision="FP32">
99
+ <dim>1</dim>
100
+ <dim>64</dim>
101
+ <dim>1</dim>
102
+ <dim>1</dim>
103
+ </port>
104
+ </output>
105
+ </layer>
106
+ <layer id="6" name="/conv1/Conv" type="Add" version="opset1">
107
+ <data auto_broadcast="numpy" />
108
+ <input>
109
+ <port id="0" precision="FP32">
110
+ <dim>-1</dim>
111
+ <dim>64</dim>
112
+ <dim>112</dim>
113
+ <dim>112</dim>
114
+ </port>
115
+ <port id="1" precision="FP32">
116
+ <dim>1</dim>
117
+ <dim>64</dim>
118
+ <dim>1</dim>
119
+ <dim>1</dim>
120
+ </port>
121
+ </input>
122
+ <output>
123
+ <port id="2" precision="FP32" names="/conv1/Conv_output_0">
124
+ <dim>-1</dim>
125
+ <dim>64</dim>
126
+ <dim>112</dim>
127
+ <dim>112</dim>
128
+ </port>
129
+ </output>
130
+ </layer>
131
+ <layer id="7" name="/relu/Relu" type="ReLU" version="opset1">
132
+ <input>
133
+ <port id="0" precision="FP32">
134
+ <dim>-1</dim>
135
+ <dim>64</dim>
136
+ <dim>112</dim>
137
+ <dim>112</dim>
138
+ </port>
139
+ </input>
140
+ <output>
141
+ <port id="1" precision="FP32" names="/relu/Relu_output_0">
142
+ <dim>-1</dim>
143
+ <dim>64</dim>
144
+ <dim>112</dim>
145
+ <dim>112</dim>
146
+ </port>
147
+ </output>
148
+ </layer>
149
+ <layer id="8" name="/maxpool/MaxPool" type="MaxPool" version="opset8">
150
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" kernel="3, 3" rounding_type="floor" auto_pad="explicit" index_element_type="i64" axis="0" />
151
+ <input>
152
+ <port id="0" precision="FP32">
153
+ <dim>-1</dim>
154
+ <dim>64</dim>
155
+ <dim>112</dim>
156
+ <dim>112</dim>
157
+ </port>
158
+ </input>
159
+ <output>
160
+ <port id="1" precision="FP32" names="/maxpool/MaxPool_output_0">
161
+ <dim>-1</dim>
162
+ <dim>64</dim>
163
+ <dim>56</dim>
164
+ <dim>56</dim>
165
+ </port>
166
+ <port id="2" precision="I64">
167
+ <dim>-1</dim>
168
+ <dim>64</dim>
169
+ <dim>56</dim>
170
+ <dim>56</dim>
171
+ </port>
172
+ </output>
173
+ </layer>
174
+ <layer id="9" name="onnx::Conv_196_compressed" type="Const" version="opset1">
175
+ <data element_type="f16" shape="64, 64, 3, 3" offset="18944" size="73728" />
176
+ <output>
177
+ <port id="0" precision="FP16" names="onnx::Conv_196">
178
+ <dim>64</dim>
179
+ <dim>64</dim>
180
+ <dim>3</dim>
181
+ <dim>3</dim>
182
+ </port>
183
+ </output>
184
+ </layer>
185
+ <layer id="10" name="onnx::Conv_196" type="Convert" version="opset1">
186
+ <data destination_type="f32" />
187
+ <rt_info>
188
+ <attribute name="decompression" version="0" />
189
+ </rt_info>
190
+ <input>
191
+ <port id="0" precision="FP16">
192
+ <dim>64</dim>
193
+ <dim>64</dim>
194
+ <dim>3</dim>
195
+ <dim>3</dim>
196
+ </port>
197
+ </input>
198
+ <output>
199
+ <port id="1" precision="FP32">
200
+ <dim>64</dim>
201
+ <dim>64</dim>
202
+ <dim>3</dim>
203
+ <dim>3</dim>
204
+ </port>
205
+ </output>
206
+ </layer>
207
+ <layer id="11" name="/layer1/layer1.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
208
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
209
+ <input>
210
+ <port id="0" precision="FP32">
211
+ <dim>-1</dim>
212
+ <dim>64</dim>
213
+ <dim>56</dim>
214
+ <dim>56</dim>
215
+ </port>
216
+ <port id="1" precision="FP32">
217
+ <dim>64</dim>
218
+ <dim>64</dim>
219
+ <dim>3</dim>
220
+ <dim>3</dim>
221
+ </port>
222
+ </input>
223
+ <output>
224
+ <port id="2" precision="FP32">
225
+ <dim>-1</dim>
226
+ <dim>64</dim>
227
+ <dim>56</dim>
228
+ <dim>56</dim>
229
+ </port>
230
+ </output>
231
+ </layer>
232
+ <layer id="12" name="Reshape_32_compressed" type="Const" version="opset1">
233
+ <data element_type="f16" shape="1, 64, 1, 1" offset="92672" size="128" />
234
+ <output>
235
+ <port id="0" precision="FP16">
236
+ <dim>1</dim>
237
+ <dim>64</dim>
238
+ <dim>1</dim>
239
+ <dim>1</dim>
240
+ </port>
241
+ </output>
242
+ </layer>
243
+ <layer id="13" name="Reshape_32" type="Convert" version="opset1">
244
+ <data destination_type="f32" />
245
+ <rt_info>
246
+ <attribute name="decompression" version="0" />
247
+ </rt_info>
248
+ <input>
249
+ <port id="0" precision="FP16">
250
+ <dim>1</dim>
251
+ <dim>64</dim>
252
+ <dim>1</dim>
253
+ <dim>1</dim>
254
+ </port>
255
+ </input>
256
+ <output>
257
+ <port id="1" precision="FP32">
258
+ <dim>1</dim>
259
+ <dim>64</dim>
260
+ <dim>1</dim>
261
+ <dim>1</dim>
262
+ </port>
263
+ </output>
264
+ </layer>
265
+ <layer id="14" name="/layer1/layer1.0/conv1/Conv" type="Add" version="opset1">
266
+ <data auto_broadcast="numpy" />
267
+ <input>
268
+ <port id="0" precision="FP32">
269
+ <dim>-1</dim>
270
+ <dim>64</dim>
271
+ <dim>56</dim>
272
+ <dim>56</dim>
273
+ </port>
274
+ <port id="1" precision="FP32">
275
+ <dim>1</dim>
276
+ <dim>64</dim>
277
+ <dim>1</dim>
278
+ <dim>1</dim>
279
+ </port>
280
+ </input>
281
+ <output>
282
+ <port id="2" precision="FP32" names="/layer1/layer1.0/conv1/Conv_output_0">
283
+ <dim>-1</dim>
284
+ <dim>64</dim>
285
+ <dim>56</dim>
286
+ <dim>56</dim>
287
+ </port>
288
+ </output>
289
+ </layer>
290
+ <layer id="15" name="/layer1/layer1.0/relu/Relu" type="ReLU" version="opset1">
291
+ <input>
292
+ <port id="0" precision="FP32">
293
+ <dim>-1</dim>
294
+ <dim>64</dim>
295
+ <dim>56</dim>
296
+ <dim>56</dim>
297
+ </port>
298
+ </input>
299
+ <output>
300
+ <port id="1" precision="FP32" names="/layer1/layer1.0/relu/Relu_output_0">
301
+ <dim>-1</dim>
302
+ <dim>64</dim>
303
+ <dim>56</dim>
304
+ <dim>56</dim>
305
+ </port>
306
+ </output>
307
+ </layer>
308
+ <layer id="16" name="onnx::Conv_199_compressed" type="Const" version="opset1">
309
+ <data element_type="f16" shape="64, 64, 3, 3" offset="92800" size="73728" />
310
+ <output>
311
+ <port id="0" precision="FP16" names="onnx::Conv_199">
312
+ <dim>64</dim>
313
+ <dim>64</dim>
314
+ <dim>3</dim>
315
+ <dim>3</dim>
316
+ </port>
317
+ </output>
318
+ </layer>
319
+ <layer id="17" name="onnx::Conv_199" type="Convert" version="opset1">
320
+ <data destination_type="f32" />
321
+ <rt_info>
322
+ <attribute name="decompression" version="0" />
323
+ </rt_info>
324
+ <input>
325
+ <port id="0" precision="FP16">
326
+ <dim>64</dim>
327
+ <dim>64</dim>
328
+ <dim>3</dim>
329
+ <dim>3</dim>
330
+ </port>
331
+ </input>
332
+ <output>
333
+ <port id="1" precision="FP32">
334
+ <dim>64</dim>
335
+ <dim>64</dim>
336
+ <dim>3</dim>
337
+ <dim>3</dim>
338
+ </port>
339
+ </output>
340
+ </layer>
341
+ <layer id="18" name="/layer1/layer1.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
342
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
343
+ <input>
344
+ <port id="0" precision="FP32">
345
+ <dim>-1</dim>
346
+ <dim>64</dim>
347
+ <dim>56</dim>
348
+ <dim>56</dim>
349
+ </port>
350
+ <port id="1" precision="FP32">
351
+ <dim>64</dim>
352
+ <dim>64</dim>
353
+ <dim>3</dim>
354
+ <dim>3</dim>
355
+ </port>
356
+ </input>
357
+ <output>
358
+ <port id="2" precision="FP32">
359
+ <dim>-1</dim>
360
+ <dim>64</dim>
361
+ <dim>56</dim>
362
+ <dim>56</dim>
363
+ </port>
364
+ </output>
365
+ </layer>
366
+ <layer id="19" name="Reshape_48_compressed" type="Const" version="opset1">
367
+ <data element_type="f16" shape="1, 64, 1, 1" offset="166528" size="128" />
368
+ <output>
369
+ <port id="0" precision="FP16">
370
+ <dim>1</dim>
371
+ <dim>64</dim>
372
+ <dim>1</dim>
373
+ <dim>1</dim>
374
+ </port>
375
+ </output>
376
+ </layer>
377
+ <layer id="20" name="Reshape_48" type="Convert" version="opset1">
378
+ <data destination_type="f32" />
379
+ <rt_info>
380
+ <attribute name="decompression" version="0" />
381
+ </rt_info>
382
+ <input>
383
+ <port id="0" precision="FP16">
384
+ <dim>1</dim>
385
+ <dim>64</dim>
386
+ <dim>1</dim>
387
+ <dim>1</dim>
388
+ </port>
389
+ </input>
390
+ <output>
391
+ <port id="1" precision="FP32">
392
+ <dim>1</dim>
393
+ <dim>64</dim>
394
+ <dim>1</dim>
395
+ <dim>1</dim>
396
+ </port>
397
+ </output>
398
+ </layer>
399
+ <layer id="21" name="/layer1/layer1.0/conv2/Conv" type="Add" version="opset1">
400
+ <data auto_broadcast="numpy" />
401
+ <input>
402
+ <port id="0" precision="FP32">
403
+ <dim>-1</dim>
404
+ <dim>64</dim>
405
+ <dim>56</dim>
406
+ <dim>56</dim>
407
+ </port>
408
+ <port id="1" precision="FP32">
409
+ <dim>1</dim>
410
+ <dim>64</dim>
411
+ <dim>1</dim>
412
+ <dim>1</dim>
413
+ </port>
414
+ </input>
415
+ <output>
416
+ <port id="2" precision="FP32" names="/layer1/layer1.0/conv2/Conv_output_0">
417
+ <dim>-1</dim>
418
+ <dim>64</dim>
419
+ <dim>56</dim>
420
+ <dim>56</dim>
421
+ </port>
422
+ </output>
423
+ </layer>
424
+ <layer id="22" name="/layer1/layer1.0/Add" type="Add" version="opset1">
425
+ <data auto_broadcast="numpy" />
426
+ <input>
427
+ <port id="0" precision="FP32">
428
+ <dim>-1</dim>
429
+ <dim>64</dim>
430
+ <dim>56</dim>
431
+ <dim>56</dim>
432
+ </port>
433
+ <port id="1" precision="FP32">
434
+ <dim>-1</dim>
435
+ <dim>64</dim>
436
+ <dim>56</dim>
437
+ <dim>56</dim>
438
+ </port>
439
+ </input>
440
+ <output>
441
+ <port id="2" precision="FP32" names="/layer1/layer1.0/Add_output_0">
442
+ <dim>-1</dim>
443
+ <dim>64</dim>
444
+ <dim>56</dim>
445
+ <dim>56</dim>
446
+ </port>
447
+ </output>
448
+ </layer>
449
+ <layer id="23" name="/layer1/layer1.0/relu_1/Relu" type="ReLU" version="opset1">
450
+ <input>
451
+ <port id="0" precision="FP32">
452
+ <dim>-1</dim>
453
+ <dim>64</dim>
454
+ <dim>56</dim>
455
+ <dim>56</dim>
456
+ </port>
457
+ </input>
458
+ <output>
459
+ <port id="1" precision="FP32" names="/layer1/layer1.0/relu_1/Relu_output_0">
460
+ <dim>-1</dim>
461
+ <dim>64</dim>
462
+ <dim>56</dim>
463
+ <dim>56</dim>
464
+ </port>
465
+ </output>
466
+ </layer>
467
+ <layer id="24" name="onnx::Conv_202_compressed" type="Const" version="opset1">
468
+ <data element_type="f16" shape="64, 64, 3, 3" offset="166656" size="73728" />
469
+ <output>
470
+ <port id="0" precision="FP16" names="onnx::Conv_202">
471
+ <dim>64</dim>
472
+ <dim>64</dim>
473
+ <dim>3</dim>
474
+ <dim>3</dim>
475
+ </port>
476
+ </output>
477
+ </layer>
478
+ <layer id="25" name="onnx::Conv_202" type="Convert" version="opset1">
479
+ <data destination_type="f32" />
480
+ <rt_info>
481
+ <attribute name="decompression" version="0" />
482
+ </rt_info>
483
+ <input>
484
+ <port id="0" precision="FP16">
485
+ <dim>64</dim>
486
+ <dim>64</dim>
487
+ <dim>3</dim>
488
+ <dim>3</dim>
489
+ </port>
490
+ </input>
491
+ <output>
492
+ <port id="1" precision="FP32">
493
+ <dim>64</dim>
494
+ <dim>64</dim>
495
+ <dim>3</dim>
496
+ <dim>3</dim>
497
+ </port>
498
+ </output>
499
+ </layer>
500
+ <layer id="26" name="/layer1/layer1.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
501
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
502
+ <input>
503
+ <port id="0" precision="FP32">
504
+ <dim>-1</dim>
505
+ <dim>64</dim>
506
+ <dim>56</dim>
507
+ <dim>56</dim>
508
+ </port>
509
+ <port id="1" precision="FP32">
510
+ <dim>64</dim>
511
+ <dim>64</dim>
512
+ <dim>3</dim>
513
+ <dim>3</dim>
514
+ </port>
515
+ </input>
516
+ <output>
517
+ <port id="2" precision="FP32">
518
+ <dim>-1</dim>
519
+ <dim>64</dim>
520
+ <dim>56</dim>
521
+ <dim>56</dim>
522
+ </port>
523
+ </output>
524
+ </layer>
525
+ <layer id="27" name="Reshape_65_compressed" type="Const" version="opset1">
526
+ <data element_type="f16" shape="1, 64, 1, 1" offset="240384" size="128" />
527
+ <output>
528
+ <port id="0" precision="FP16">
529
+ <dim>1</dim>
530
+ <dim>64</dim>
531
+ <dim>1</dim>
532
+ <dim>1</dim>
533
+ </port>
534
+ </output>
535
+ </layer>
536
+ <layer id="28" name="Reshape_65" type="Convert" version="opset1">
537
+ <data destination_type="f32" />
538
+ <rt_info>
539
+ <attribute name="decompression" version="0" />
540
+ </rt_info>
541
+ <input>
542
+ <port id="0" precision="FP16">
543
+ <dim>1</dim>
544
+ <dim>64</dim>
545
+ <dim>1</dim>
546
+ <dim>1</dim>
547
+ </port>
548
+ </input>
549
+ <output>
550
+ <port id="1" precision="FP32">
551
+ <dim>1</dim>
552
+ <dim>64</dim>
553
+ <dim>1</dim>
554
+ <dim>1</dim>
555
+ </port>
556
+ </output>
557
+ </layer>
558
+ <layer id="29" name="/layer1/layer1.1/conv1/Conv" type="Add" version="opset1">
559
+ <data auto_broadcast="numpy" />
560
+ <input>
561
+ <port id="0" precision="FP32">
562
+ <dim>-1</dim>
563
+ <dim>64</dim>
564
+ <dim>56</dim>
565
+ <dim>56</dim>
566
+ </port>
567
+ <port id="1" precision="FP32">
568
+ <dim>1</dim>
569
+ <dim>64</dim>
570
+ <dim>1</dim>
571
+ <dim>1</dim>
572
+ </port>
573
+ </input>
574
+ <output>
575
+ <port id="2" precision="FP32" names="/layer1/layer1.1/conv1/Conv_output_0">
576
+ <dim>-1</dim>
577
+ <dim>64</dim>
578
+ <dim>56</dim>
579
+ <dim>56</dim>
580
+ </port>
581
+ </output>
582
+ </layer>
583
+ <layer id="30" name="/layer1/layer1.1/relu/Relu" type="ReLU" version="opset1">
584
+ <input>
585
+ <port id="0" precision="FP32">
586
+ <dim>-1</dim>
587
+ <dim>64</dim>
588
+ <dim>56</dim>
589
+ <dim>56</dim>
590
+ </port>
591
+ </input>
592
+ <output>
593
+ <port id="1" precision="FP32" names="/layer1/layer1.1/relu/Relu_output_0">
594
+ <dim>-1</dim>
595
+ <dim>64</dim>
596
+ <dim>56</dim>
597
+ <dim>56</dim>
598
+ </port>
599
+ </output>
600
+ </layer>
601
+ <layer id="31" name="onnx::Conv_205_compressed" type="Const" version="opset1">
602
+ <data element_type="f16" shape="64, 64, 3, 3" offset="240512" size="73728" />
603
+ <output>
604
+ <port id="0" precision="FP16" names="onnx::Conv_205">
605
+ <dim>64</dim>
606
+ <dim>64</dim>
607
+ <dim>3</dim>
608
+ <dim>3</dim>
609
+ </port>
610
+ </output>
611
+ </layer>
612
+ <layer id="32" name="onnx::Conv_205" type="Convert" version="opset1">
613
+ <data destination_type="f32" />
614
+ <rt_info>
615
+ <attribute name="decompression" version="0" />
616
+ </rt_info>
617
+ <input>
618
+ <port id="0" precision="FP16">
619
+ <dim>64</dim>
620
+ <dim>64</dim>
621
+ <dim>3</dim>
622
+ <dim>3</dim>
623
+ </port>
624
+ </input>
625
+ <output>
626
+ <port id="1" precision="FP32">
627
+ <dim>64</dim>
628
+ <dim>64</dim>
629
+ <dim>3</dim>
630
+ <dim>3</dim>
631
+ </port>
632
+ </output>
633
+ </layer>
634
+ <layer id="33" name="/layer1/layer1.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
635
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
636
+ <input>
637
+ <port id="0" precision="FP32">
638
+ <dim>-1</dim>
639
+ <dim>64</dim>
640
+ <dim>56</dim>
641
+ <dim>56</dim>
642
+ </port>
643
+ <port id="1" precision="FP32">
644
+ <dim>64</dim>
645
+ <dim>64</dim>
646
+ <dim>3</dim>
647
+ <dim>3</dim>
648
+ </port>
649
+ </input>
650
+ <output>
651
+ <port id="2" precision="FP32">
652
+ <dim>-1</dim>
653
+ <dim>64</dim>
654
+ <dim>56</dim>
655
+ <dim>56</dim>
656
+ </port>
657
+ </output>
658
+ </layer>
659
+ <layer id="34" name="Reshape_81_compressed" type="Const" version="opset1">
660
+ <data element_type="f16" shape="1, 64, 1, 1" offset="314240" size="128" />
661
+ <output>
662
+ <port id="0" precision="FP16">
663
+ <dim>1</dim>
664
+ <dim>64</dim>
665
+ <dim>1</dim>
666
+ <dim>1</dim>
667
+ </port>
668
+ </output>
669
+ </layer>
670
+ <layer id="35" name="Reshape_81" type="Convert" version="opset1">
671
+ <data destination_type="f32" />
672
+ <rt_info>
673
+ <attribute name="decompression" version="0" />
674
+ </rt_info>
675
+ <input>
676
+ <port id="0" precision="FP16">
677
+ <dim>1</dim>
678
+ <dim>64</dim>
679
+ <dim>1</dim>
680
+ <dim>1</dim>
681
+ </port>
682
+ </input>
683
+ <output>
684
+ <port id="1" precision="FP32">
685
+ <dim>1</dim>
686
+ <dim>64</dim>
687
+ <dim>1</dim>
688
+ <dim>1</dim>
689
+ </port>
690
+ </output>
691
+ </layer>
692
+ <layer id="36" name="/layer1/layer1.1/conv2/Conv" type="Add" version="opset1">
693
+ <data auto_broadcast="numpy" />
694
+ <input>
695
+ <port id="0" precision="FP32">
696
+ <dim>-1</dim>
697
+ <dim>64</dim>
698
+ <dim>56</dim>
699
+ <dim>56</dim>
700
+ </port>
701
+ <port id="1" precision="FP32">
702
+ <dim>1</dim>
703
+ <dim>64</dim>
704
+ <dim>1</dim>
705
+ <dim>1</dim>
706
+ </port>
707
+ </input>
708
+ <output>
709
+ <port id="2" precision="FP32" names="/layer1/layer1.1/conv2/Conv_output_0">
710
+ <dim>-1</dim>
711
+ <dim>64</dim>
712
+ <dim>56</dim>
713
+ <dim>56</dim>
714
+ </port>
715
+ </output>
716
+ </layer>
717
+ <layer id="37" name="/layer1/layer1.1/Add" type="Add" version="opset1">
718
+ <data auto_broadcast="numpy" />
719
+ <input>
720
+ <port id="0" precision="FP32">
721
+ <dim>-1</dim>
722
+ <dim>64</dim>
723
+ <dim>56</dim>
724
+ <dim>56</dim>
725
+ </port>
726
+ <port id="1" precision="FP32">
727
+ <dim>-1</dim>
728
+ <dim>64</dim>
729
+ <dim>56</dim>
730
+ <dim>56</dim>
731
+ </port>
732
+ </input>
733
+ <output>
734
+ <port id="2" precision="FP32" names="/layer1/layer1.1/Add_output_0">
735
+ <dim>-1</dim>
736
+ <dim>64</dim>
737
+ <dim>56</dim>
738
+ <dim>56</dim>
739
+ </port>
740
+ </output>
741
+ </layer>
742
+ <layer id="38" name="/layer1/layer1.1/relu_1/Relu" type="ReLU" version="opset1">
743
+ <input>
744
+ <port id="0" precision="FP32">
745
+ <dim>-1</dim>
746
+ <dim>64</dim>
747
+ <dim>56</dim>
748
+ <dim>56</dim>
749
+ </port>
750
+ </input>
751
+ <output>
752
+ <port id="1" precision="FP32" names="/layer1/layer1.1/relu_1/Relu_output_0">
753
+ <dim>-1</dim>
754
+ <dim>64</dim>
755
+ <dim>56</dim>
756
+ <dim>56</dim>
757
+ </port>
758
+ </output>
759
+ </layer>
760
+ <layer id="39" name="onnx::Conv_208_compressed" type="Const" version="opset1">
761
+ <data element_type="f16" shape="128, 64, 3, 3" offset="314368" size="147456" />
762
+ <output>
763
+ <port id="0" precision="FP16" names="onnx::Conv_208">
764
+ <dim>128</dim>
765
+ <dim>64</dim>
766
+ <dim>3</dim>
767
+ <dim>3</dim>
768
+ </port>
769
+ </output>
770
+ </layer>
771
+ <layer id="40" name="onnx::Conv_208" type="Convert" version="opset1">
772
+ <data destination_type="f32" />
773
+ <rt_info>
774
+ <attribute name="decompression" version="0" />
775
+ </rt_info>
776
+ <input>
777
+ <port id="0" precision="FP16">
778
+ <dim>128</dim>
779
+ <dim>64</dim>
780
+ <dim>3</dim>
781
+ <dim>3</dim>
782
+ </port>
783
+ </input>
784
+ <output>
785
+ <port id="1" precision="FP32">
786
+ <dim>128</dim>
787
+ <dim>64</dim>
788
+ <dim>3</dim>
789
+ <dim>3</dim>
790
+ </port>
791
+ </output>
792
+ </layer>
793
+ <layer id="41" name="/layer2/layer2.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
794
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
795
+ <input>
796
+ <port id="0" precision="FP32">
797
+ <dim>-1</dim>
798
+ <dim>64</dim>
799
+ <dim>56</dim>
800
+ <dim>56</dim>
801
+ </port>
802
+ <port id="1" precision="FP32">
803
+ <dim>128</dim>
804
+ <dim>64</dim>
805
+ <dim>3</dim>
806
+ <dim>3</dim>
807
+ </port>
808
+ </input>
809
+ <output>
810
+ <port id="2" precision="FP32">
811
+ <dim>-1</dim>
812
+ <dim>128</dim>
813
+ <dim>28</dim>
814
+ <dim>28</dim>
815
+ </port>
816
+ </output>
817
+ </layer>
818
+ <layer id="42" name="Reshape_98_compressed" type="Const" version="opset1">
819
+ <data element_type="f16" shape="1, 128, 1, 1" offset="461824" size="256" />
820
+ <output>
821
+ <port id="0" precision="FP16">
822
+ <dim>1</dim>
823
+ <dim>128</dim>
824
+ <dim>1</dim>
825
+ <dim>1</dim>
826
+ </port>
827
+ </output>
828
+ </layer>
829
+ <layer id="43" name="Reshape_98" type="Convert" version="opset1">
830
+ <data destination_type="f32" />
831
+ <rt_info>
832
+ <attribute name="decompression" version="0" />
833
+ </rt_info>
834
+ <input>
835
+ <port id="0" precision="FP16">
836
+ <dim>1</dim>
837
+ <dim>128</dim>
838
+ <dim>1</dim>
839
+ <dim>1</dim>
840
+ </port>
841
+ </input>
842
+ <output>
843
+ <port id="1" precision="FP32">
844
+ <dim>1</dim>
845
+ <dim>128</dim>
846
+ <dim>1</dim>
847
+ <dim>1</dim>
848
+ </port>
849
+ </output>
850
+ </layer>
851
+ <layer id="44" name="/layer2/layer2.0/conv1/Conv" type="Add" version="opset1">
852
+ <data auto_broadcast="numpy" />
853
+ <input>
854
+ <port id="0" precision="FP32">
855
+ <dim>-1</dim>
856
+ <dim>128</dim>
857
+ <dim>28</dim>
858
+ <dim>28</dim>
859
+ </port>
860
+ <port id="1" precision="FP32">
861
+ <dim>1</dim>
862
+ <dim>128</dim>
863
+ <dim>1</dim>
864
+ <dim>1</dim>
865
+ </port>
866
+ </input>
867
+ <output>
868
+ <port id="2" precision="FP32" names="/layer2/layer2.0/conv1/Conv_output_0">
869
+ <dim>-1</dim>
870
+ <dim>128</dim>
871
+ <dim>28</dim>
872
+ <dim>28</dim>
873
+ </port>
874
+ </output>
875
+ </layer>
876
+ <layer id="45" name="/layer2/layer2.0/relu/Relu" type="ReLU" version="opset1">
877
+ <input>
878
+ <port id="0" precision="FP32">
879
+ <dim>-1</dim>
880
+ <dim>128</dim>
881
+ <dim>28</dim>
882
+ <dim>28</dim>
883
+ </port>
884
+ </input>
885
+ <output>
886
+ <port id="1" precision="FP32" names="/layer2/layer2.0/relu/Relu_output_0">
887
+ <dim>-1</dim>
888
+ <dim>128</dim>
889
+ <dim>28</dim>
890
+ <dim>28</dim>
891
+ </port>
892
+ </output>
893
+ </layer>
894
+ <layer id="46" name="onnx::Conv_211_compressed" type="Const" version="opset1">
895
+ <data element_type="f16" shape="128, 128, 3, 3" offset="462080" size="294912" />
896
+ <output>
897
+ <port id="0" precision="FP16" names="onnx::Conv_211">
898
+ <dim>128</dim>
899
+ <dim>128</dim>
900
+ <dim>3</dim>
901
+ <dim>3</dim>
902
+ </port>
903
+ </output>
904
+ </layer>
905
+ <layer id="47" name="onnx::Conv_211" type="Convert" version="opset1">
906
+ <data destination_type="f32" />
907
+ <rt_info>
908
+ <attribute name="decompression" version="0" />
909
+ </rt_info>
910
+ <input>
911
+ <port id="0" precision="FP16">
912
+ <dim>128</dim>
913
+ <dim>128</dim>
914
+ <dim>3</dim>
915
+ <dim>3</dim>
916
+ </port>
917
+ </input>
918
+ <output>
919
+ <port id="1" precision="FP32">
920
+ <dim>128</dim>
921
+ <dim>128</dim>
922
+ <dim>3</dim>
923
+ <dim>3</dim>
924
+ </port>
925
+ </output>
926
+ </layer>
927
+ <layer id="48" name="/layer2/layer2.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
928
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
929
+ <input>
930
+ <port id="0" precision="FP32">
931
+ <dim>-1</dim>
932
+ <dim>128</dim>
933
+ <dim>28</dim>
934
+ <dim>28</dim>
935
+ </port>
936
+ <port id="1" precision="FP32">
937
+ <dim>128</dim>
938
+ <dim>128</dim>
939
+ <dim>3</dim>
940
+ <dim>3</dim>
941
+ </port>
942
+ </input>
943
+ <output>
944
+ <port id="2" precision="FP32">
945
+ <dim>-1</dim>
946
+ <dim>128</dim>
947
+ <dim>28</dim>
948
+ <dim>28</dim>
949
+ </port>
950
+ </output>
951
+ </layer>
952
+ <layer id="49" name="Reshape_129_compressed" type="Const" version="opset1">
953
+ <data element_type="f16" shape="1, 128, 1, 1" offset="756992" size="256" />
954
+ <output>
955
+ <port id="0" precision="FP16">
956
+ <dim>1</dim>
957
+ <dim>128</dim>
958
+ <dim>1</dim>
959
+ <dim>1</dim>
960
+ </port>
961
+ </output>
962
+ </layer>
963
+ <layer id="50" name="Reshape_129" type="Convert" version="opset1">
964
+ <data destination_type="f32" />
965
+ <rt_info>
966
+ <attribute name="decompression" version="0" />
967
+ </rt_info>
968
+ <input>
969
+ <port id="0" precision="FP16">
970
+ <dim>1</dim>
971
+ <dim>128</dim>
972
+ <dim>1</dim>
973
+ <dim>1</dim>
974
+ </port>
975
+ </input>
976
+ <output>
977
+ <port id="1" precision="FP32">
978
+ <dim>1</dim>
979
+ <dim>128</dim>
980
+ <dim>1</dim>
981
+ <dim>1</dim>
982
+ </port>
983
+ </output>
984
+ </layer>
985
+ <layer id="51" name="/layer2/layer2.0/conv2/Conv" type="Add" version="opset1">
986
+ <data auto_broadcast="numpy" />
987
+ <input>
988
+ <port id="0" precision="FP32">
989
+ <dim>-1</dim>
990
+ <dim>128</dim>
991
+ <dim>28</dim>
992
+ <dim>28</dim>
993
+ </port>
994
+ <port id="1" precision="FP32">
995
+ <dim>1</dim>
996
+ <dim>128</dim>
997
+ <dim>1</dim>
998
+ <dim>1</dim>
999
+ </port>
1000
+ </input>
1001
+ <output>
1002
+ <port id="2" precision="FP32" names="/layer2/layer2.0/conv2/Conv_output_0">
1003
+ <dim>-1</dim>
1004
+ <dim>128</dim>
1005
+ <dim>28</dim>
1006
+ <dim>28</dim>
1007
+ </port>
1008
+ </output>
1009
+ </layer>
1010
+ <layer id="52" name="onnx::Conv_214_compressed" type="Const" version="opset1">
1011
+ <data element_type="f16" shape="128, 64, 1, 1" offset="757248" size="16384" />
1012
+ <output>
1013
+ <port id="0" precision="FP16" names="onnx::Conv_214">
1014
+ <dim>128</dim>
1015
+ <dim>64</dim>
1016
+ <dim>1</dim>
1017
+ <dim>1</dim>
1018
+ </port>
1019
+ </output>
1020
+ </layer>
1021
+ <layer id="53" name="onnx::Conv_214" type="Convert" version="opset1">
1022
+ <data destination_type="f32" />
1023
+ <rt_info>
1024
+ <attribute name="decompression" version="0" />
1025
+ </rt_info>
1026
+ <input>
1027
+ <port id="0" precision="FP16">
1028
+ <dim>128</dim>
1029
+ <dim>64</dim>
1030
+ <dim>1</dim>
1031
+ <dim>1</dim>
1032
+ </port>
1033
+ </input>
1034
+ <output>
1035
+ <port id="1" precision="FP32">
1036
+ <dim>128</dim>
1037
+ <dim>64</dim>
1038
+ <dim>1</dim>
1039
+ <dim>1</dim>
1040
+ </port>
1041
+ </output>
1042
+ </layer>
1043
+ <layer id="54" name="/layer2/layer2.0/downsample/downsample.0/Conv/WithoutBiases" type="Convolution" version="opset1">
1044
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
1045
+ <input>
1046
+ <port id="0" precision="FP32">
1047
+ <dim>-1</dim>
1048
+ <dim>64</dim>
1049
+ <dim>56</dim>
1050
+ <dim>56</dim>
1051
+ </port>
1052
+ <port id="1" precision="FP32">
1053
+ <dim>128</dim>
1054
+ <dim>64</dim>
1055
+ <dim>1</dim>
1056
+ <dim>1</dim>
1057
+ </port>
1058
+ </input>
1059
+ <output>
1060
+ <port id="2" precision="FP32">
1061
+ <dim>-1</dim>
1062
+ <dim>128</dim>
1063
+ <dim>28</dim>
1064
+ <dim>28</dim>
1065
+ </port>
1066
+ </output>
1067
+ </layer>
1068
+ <layer id="55" name="Reshape_113_compressed" type="Const" version="opset1">
1069
+ <data element_type="f16" shape="1, 128, 1, 1" offset="773632" size="256" />
1070
+ <output>
1071
+ <port id="0" precision="FP16">
1072
+ <dim>1</dim>
1073
+ <dim>128</dim>
1074
+ <dim>1</dim>
1075
+ <dim>1</dim>
1076
+ </port>
1077
+ </output>
1078
+ </layer>
1079
+ <layer id="56" name="Reshape_113" type="Convert" version="opset1">
1080
+ <data destination_type="f32" />
1081
+ <rt_info>
1082
+ <attribute name="decompression" version="0" />
1083
+ </rt_info>
1084
+ <input>
1085
+ <port id="0" precision="FP16">
1086
+ <dim>1</dim>
1087
+ <dim>128</dim>
1088
+ <dim>1</dim>
1089
+ <dim>1</dim>
1090
+ </port>
1091
+ </input>
1092
+ <output>
1093
+ <port id="1" precision="FP32">
1094
+ <dim>1</dim>
1095
+ <dim>128</dim>
1096
+ <dim>1</dim>
1097
+ <dim>1</dim>
1098
+ </port>
1099
+ </output>
1100
+ </layer>
1101
+ <layer id="57" name="/layer2/layer2.0/downsample/downsample.0/Conv" type="Add" version="opset1">
1102
+ <data auto_broadcast="numpy" />
1103
+ <input>
1104
+ <port id="0" precision="FP32">
1105
+ <dim>-1</dim>
1106
+ <dim>128</dim>
1107
+ <dim>28</dim>
1108
+ <dim>28</dim>
1109
+ </port>
1110
+ <port id="1" precision="FP32">
1111
+ <dim>1</dim>
1112
+ <dim>128</dim>
1113
+ <dim>1</dim>
1114
+ <dim>1</dim>
1115
+ </port>
1116
+ </input>
1117
+ <output>
1118
+ <port id="2" precision="FP32" names="/layer2/layer2.0/downsample/downsample.0/Conv_output_0">
1119
+ <dim>-1</dim>
1120
+ <dim>128</dim>
1121
+ <dim>28</dim>
1122
+ <dim>28</dim>
1123
+ </port>
1124
+ </output>
1125
+ </layer>
1126
+ <layer id="58" name="/layer2/layer2.0/Add" type="Add" version="opset1">
1127
+ <data auto_broadcast="numpy" />
1128
+ <input>
1129
+ <port id="0" precision="FP32">
1130
+ <dim>-1</dim>
1131
+ <dim>128</dim>
1132
+ <dim>28</dim>
1133
+ <dim>28</dim>
1134
+ </port>
1135
+ <port id="1" precision="FP32">
1136
+ <dim>-1</dim>
1137
+ <dim>128</dim>
1138
+ <dim>28</dim>
1139
+ <dim>28</dim>
1140
+ </port>
1141
+ </input>
1142
+ <output>
1143
+ <port id="2" precision="FP32" names="/layer2/layer2.0/Add_output_0">
1144
+ <dim>-1</dim>
1145
+ <dim>128</dim>
1146
+ <dim>28</dim>
1147
+ <dim>28</dim>
1148
+ </port>
1149
+ </output>
1150
+ </layer>
1151
+ <layer id="59" name="/layer2/layer2.0/relu_1/Relu" type="ReLU" version="opset1">
1152
+ <input>
1153
+ <port id="0" precision="FP32">
1154
+ <dim>-1</dim>
1155
+ <dim>128</dim>
1156
+ <dim>28</dim>
1157
+ <dim>28</dim>
1158
+ </port>
1159
+ </input>
1160
+ <output>
1161
+ <port id="1" precision="FP32" names="/layer2/layer2.0/relu_1/Relu_output_0">
1162
+ <dim>-1</dim>
1163
+ <dim>128</dim>
1164
+ <dim>28</dim>
1165
+ <dim>28</dim>
1166
+ </port>
1167
+ </output>
1168
+ </layer>
1169
+ <layer id="60" name="onnx::Conv_217_compressed" type="Const" version="opset1">
1170
+ <data element_type="f16" shape="128, 128, 3, 3" offset="773888" size="294912" />
1171
+ <output>
1172
+ <port id="0" precision="FP16" names="onnx::Conv_217">
1173
+ <dim>128</dim>
1174
+ <dim>128</dim>
1175
+ <dim>3</dim>
1176
+ <dim>3</dim>
1177
+ </port>
1178
+ </output>
1179
+ </layer>
1180
+ <layer id="61" name="onnx::Conv_217" type="Convert" version="opset1">
1181
+ <data destination_type="f32" />
1182
+ <rt_info>
1183
+ <attribute name="decompression" version="0" />
1184
+ </rt_info>
1185
+ <input>
1186
+ <port id="0" precision="FP16">
1187
+ <dim>128</dim>
1188
+ <dim>128</dim>
1189
+ <dim>3</dim>
1190
+ <dim>3</dim>
1191
+ </port>
1192
+ </input>
1193
+ <output>
1194
+ <port id="1" precision="FP32">
1195
+ <dim>128</dim>
1196
+ <dim>128</dim>
1197
+ <dim>3</dim>
1198
+ <dim>3</dim>
1199
+ </port>
1200
+ </output>
1201
+ </layer>
1202
+ <layer id="62" name="/layer2/layer2.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
1203
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1204
+ <input>
1205
+ <port id="0" precision="FP32">
1206
+ <dim>-1</dim>
1207
+ <dim>128</dim>
1208
+ <dim>28</dim>
1209
+ <dim>28</dim>
1210
+ </port>
1211
+ <port id="1" precision="FP32">
1212
+ <dim>128</dim>
1213
+ <dim>128</dim>
1214
+ <dim>3</dim>
1215
+ <dim>3</dim>
1216
+ </port>
1217
+ </input>
1218
+ <output>
1219
+ <port id="2" precision="FP32">
1220
+ <dim>-1</dim>
1221
+ <dim>128</dim>
1222
+ <dim>28</dim>
1223
+ <dim>28</dim>
1224
+ </port>
1225
+ </output>
1226
+ </layer>
1227
+ <layer id="63" name="Reshape_146_compressed" type="Const" version="opset1">
1228
+ <data element_type="f16" shape="1, 128, 1, 1" offset="1068800" size="256" />
1229
+ <output>
1230
+ <port id="0" precision="FP16">
1231
+ <dim>1</dim>
1232
+ <dim>128</dim>
1233
+ <dim>1</dim>
1234
+ <dim>1</dim>
1235
+ </port>
1236
+ </output>
1237
+ </layer>
1238
+ <layer id="64" name="Reshape_146" type="Convert" version="opset1">
1239
+ <data destination_type="f32" />
1240
+ <rt_info>
1241
+ <attribute name="decompression" version="0" />
1242
+ </rt_info>
1243
+ <input>
1244
+ <port id="0" precision="FP16">
1245
+ <dim>1</dim>
1246
+ <dim>128</dim>
1247
+ <dim>1</dim>
1248
+ <dim>1</dim>
1249
+ </port>
1250
+ </input>
1251
+ <output>
1252
+ <port id="1" precision="FP32">
1253
+ <dim>1</dim>
1254
+ <dim>128</dim>
1255
+ <dim>1</dim>
1256
+ <dim>1</dim>
1257
+ </port>
1258
+ </output>
1259
+ </layer>
1260
+ <layer id="65" name="/layer2/layer2.1/conv1/Conv" type="Add" version="opset1">
1261
+ <data auto_broadcast="numpy" />
1262
+ <input>
1263
+ <port id="0" precision="FP32">
1264
+ <dim>-1</dim>
1265
+ <dim>128</dim>
1266
+ <dim>28</dim>
1267
+ <dim>28</dim>
1268
+ </port>
1269
+ <port id="1" precision="FP32">
1270
+ <dim>1</dim>
1271
+ <dim>128</dim>
1272
+ <dim>1</dim>
1273
+ <dim>1</dim>
1274
+ </port>
1275
+ </input>
1276
+ <output>
1277
+ <port id="2" precision="FP32" names="/layer2/layer2.1/conv1/Conv_output_0">
1278
+ <dim>-1</dim>
1279
+ <dim>128</dim>
1280
+ <dim>28</dim>
1281
+ <dim>28</dim>
1282
+ </port>
1283
+ </output>
1284
+ </layer>
1285
+ <layer id="66" name="/layer2/layer2.1/relu/Relu" type="ReLU" version="opset1">
1286
+ <input>
1287
+ <port id="0" precision="FP32">
1288
+ <dim>-1</dim>
1289
+ <dim>128</dim>
1290
+ <dim>28</dim>
1291
+ <dim>28</dim>
1292
+ </port>
1293
+ </input>
1294
+ <output>
1295
+ <port id="1" precision="FP32" names="/layer2/layer2.1/relu/Relu_output_0">
1296
+ <dim>-1</dim>
1297
+ <dim>128</dim>
1298
+ <dim>28</dim>
1299
+ <dim>28</dim>
1300
+ </port>
1301
+ </output>
1302
+ </layer>
1303
+ <layer id="67" name="onnx::Conv_220_compressed" type="Const" version="opset1">
1304
+ <data element_type="f16" shape="128, 128, 3, 3" offset="1069056" size="294912" />
1305
+ <output>
1306
+ <port id="0" precision="FP16" names="onnx::Conv_220">
1307
+ <dim>128</dim>
1308
+ <dim>128</dim>
1309
+ <dim>3</dim>
1310
+ <dim>3</dim>
1311
+ </port>
1312
+ </output>
1313
+ </layer>
1314
+ <layer id="68" name="onnx::Conv_220" type="Convert" version="opset1">
1315
+ <data destination_type="f32" />
1316
+ <rt_info>
1317
+ <attribute name="decompression" version="0" />
1318
+ </rt_info>
1319
+ <input>
1320
+ <port id="0" precision="FP16">
1321
+ <dim>128</dim>
1322
+ <dim>128</dim>
1323
+ <dim>3</dim>
1324
+ <dim>3</dim>
1325
+ </port>
1326
+ </input>
1327
+ <output>
1328
+ <port id="1" precision="FP32">
1329
+ <dim>128</dim>
1330
+ <dim>128</dim>
1331
+ <dim>3</dim>
1332
+ <dim>3</dim>
1333
+ </port>
1334
+ </output>
1335
+ </layer>
1336
+ <layer id="69" name="/layer2/layer2.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
1337
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1338
+ <input>
1339
+ <port id="0" precision="FP32">
1340
+ <dim>-1</dim>
1341
+ <dim>128</dim>
1342
+ <dim>28</dim>
1343
+ <dim>28</dim>
1344
+ </port>
1345
+ <port id="1" precision="FP32">
1346
+ <dim>128</dim>
1347
+ <dim>128</dim>
1348
+ <dim>3</dim>
1349
+ <dim>3</dim>
1350
+ </port>
1351
+ </input>
1352
+ <output>
1353
+ <port id="2" precision="FP32">
1354
+ <dim>-1</dim>
1355
+ <dim>128</dim>
1356
+ <dim>28</dim>
1357
+ <dim>28</dim>
1358
+ </port>
1359
+ </output>
1360
+ </layer>
1361
+ <layer id="70" name="Reshape_162_compressed" type="Const" version="opset1">
1362
+ <data element_type="f16" shape="1, 128, 1, 1" offset="1363968" size="256" />
1363
+ <output>
1364
+ <port id="0" precision="FP16">
1365
+ <dim>1</dim>
1366
+ <dim>128</dim>
1367
+ <dim>1</dim>
1368
+ <dim>1</dim>
1369
+ </port>
1370
+ </output>
1371
+ </layer>
1372
+ <layer id="71" name="Reshape_162" type="Convert" version="opset1">
1373
+ <data destination_type="f32" />
1374
+ <rt_info>
1375
+ <attribute name="decompression" version="0" />
1376
+ </rt_info>
1377
+ <input>
1378
+ <port id="0" precision="FP16">
1379
+ <dim>1</dim>
1380
+ <dim>128</dim>
1381
+ <dim>1</dim>
1382
+ <dim>1</dim>
1383
+ </port>
1384
+ </input>
1385
+ <output>
1386
+ <port id="1" precision="FP32">
1387
+ <dim>1</dim>
1388
+ <dim>128</dim>
1389
+ <dim>1</dim>
1390
+ <dim>1</dim>
1391
+ </port>
1392
+ </output>
1393
+ </layer>
1394
+ <layer id="72" name="/layer2/layer2.1/conv2/Conv" type="Add" version="opset1">
1395
+ <data auto_broadcast="numpy" />
1396
+ <input>
1397
+ <port id="0" precision="FP32">
1398
+ <dim>-1</dim>
1399
+ <dim>128</dim>
1400
+ <dim>28</dim>
1401
+ <dim>28</dim>
1402
+ </port>
1403
+ <port id="1" precision="FP32">
1404
+ <dim>1</dim>
1405
+ <dim>128</dim>
1406
+ <dim>1</dim>
1407
+ <dim>1</dim>
1408
+ </port>
1409
+ </input>
1410
+ <output>
1411
+ <port id="2" precision="FP32" names="/layer2/layer2.1/conv2/Conv_output_0">
1412
+ <dim>-1</dim>
1413
+ <dim>128</dim>
1414
+ <dim>28</dim>
1415
+ <dim>28</dim>
1416
+ </port>
1417
+ </output>
1418
+ </layer>
1419
+ <layer id="73" name="/layer2/layer2.1/Add" type="Add" version="opset1">
1420
+ <data auto_broadcast="numpy" />
1421
+ <input>
1422
+ <port id="0" precision="FP32">
1423
+ <dim>-1</dim>
1424
+ <dim>128</dim>
1425
+ <dim>28</dim>
1426
+ <dim>28</dim>
1427
+ </port>
1428
+ <port id="1" precision="FP32">
1429
+ <dim>-1</dim>
1430
+ <dim>128</dim>
1431
+ <dim>28</dim>
1432
+ <dim>28</dim>
1433
+ </port>
1434
+ </input>
1435
+ <output>
1436
+ <port id="2" precision="FP32" names="/layer2/layer2.1/Add_output_0">
1437
+ <dim>-1</dim>
1438
+ <dim>128</dim>
1439
+ <dim>28</dim>
1440
+ <dim>28</dim>
1441
+ </port>
1442
+ </output>
1443
+ </layer>
1444
+ <layer id="74" name="/layer2/layer2.1/relu_1/Relu" type="ReLU" version="opset1">
1445
+ <input>
1446
+ <port id="0" precision="FP32">
1447
+ <dim>-1</dim>
1448
+ <dim>128</dim>
1449
+ <dim>28</dim>
1450
+ <dim>28</dim>
1451
+ </port>
1452
+ </input>
1453
+ <output>
1454
+ <port id="1" precision="FP32" names="/layer2/layer2.1/relu_1/Relu_output_0">
1455
+ <dim>-1</dim>
1456
+ <dim>128</dim>
1457
+ <dim>28</dim>
1458
+ <dim>28</dim>
1459
+ </port>
1460
+ </output>
1461
+ </layer>
1462
+ <layer id="75" name="onnx::Conv_223_compressed" type="Const" version="opset1">
1463
+ <data element_type="f16" shape="256, 128, 3, 3" offset="1364224" size="589824" />
1464
+ <output>
1465
+ <port id="0" precision="FP16" names="onnx::Conv_223">
1466
+ <dim>256</dim>
1467
+ <dim>128</dim>
1468
+ <dim>3</dim>
1469
+ <dim>3</dim>
1470
+ </port>
1471
+ </output>
1472
+ </layer>
1473
+ <layer id="76" name="onnx::Conv_223" type="Convert" version="opset1">
1474
+ <data destination_type="f32" />
1475
+ <rt_info>
1476
+ <attribute name="decompression" version="0" />
1477
+ </rt_info>
1478
+ <input>
1479
+ <port id="0" precision="FP16">
1480
+ <dim>256</dim>
1481
+ <dim>128</dim>
1482
+ <dim>3</dim>
1483
+ <dim>3</dim>
1484
+ </port>
1485
+ </input>
1486
+ <output>
1487
+ <port id="1" precision="FP32">
1488
+ <dim>256</dim>
1489
+ <dim>128</dim>
1490
+ <dim>3</dim>
1491
+ <dim>3</dim>
1492
+ </port>
1493
+ </output>
1494
+ </layer>
1495
+ <layer id="77" name="/layer3/layer3.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
1496
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1497
+ <input>
1498
+ <port id="0" precision="FP32">
1499
+ <dim>-1</dim>
1500
+ <dim>128</dim>
1501
+ <dim>28</dim>
1502
+ <dim>28</dim>
1503
+ </port>
1504
+ <port id="1" precision="FP32">
1505
+ <dim>256</dim>
1506
+ <dim>128</dim>
1507
+ <dim>3</dim>
1508
+ <dim>3</dim>
1509
+ </port>
1510
+ </input>
1511
+ <output>
1512
+ <port id="2" precision="FP32">
1513
+ <dim>-1</dim>
1514
+ <dim>256</dim>
1515
+ <dim>14</dim>
1516
+ <dim>14</dim>
1517
+ </port>
1518
+ </output>
1519
+ </layer>
1520
+ <layer id="78" name="Reshape_179_compressed" type="Const" version="opset1">
1521
+ <data element_type="f16" shape="1, 256, 1, 1" offset="1954048" size="512" />
1522
+ <output>
1523
+ <port id="0" precision="FP16">
1524
+ <dim>1</dim>
1525
+ <dim>256</dim>
1526
+ <dim>1</dim>
1527
+ <dim>1</dim>
1528
+ </port>
1529
+ </output>
1530
+ </layer>
1531
+ <layer id="79" name="Reshape_179" type="Convert" version="opset1">
1532
+ <data destination_type="f32" />
1533
+ <rt_info>
1534
+ <attribute name="decompression" version="0" />
1535
+ </rt_info>
1536
+ <input>
1537
+ <port id="0" precision="FP16">
1538
+ <dim>1</dim>
1539
+ <dim>256</dim>
1540
+ <dim>1</dim>
1541
+ <dim>1</dim>
1542
+ </port>
1543
+ </input>
1544
+ <output>
1545
+ <port id="1" precision="FP32">
1546
+ <dim>1</dim>
1547
+ <dim>256</dim>
1548
+ <dim>1</dim>
1549
+ <dim>1</dim>
1550
+ </port>
1551
+ </output>
1552
+ </layer>
1553
+ <layer id="80" name="/layer3/layer3.0/conv1/Conv" type="Add" version="opset1">
1554
+ <data auto_broadcast="numpy" />
1555
+ <input>
1556
+ <port id="0" precision="FP32">
1557
+ <dim>-1</dim>
1558
+ <dim>256</dim>
1559
+ <dim>14</dim>
1560
+ <dim>14</dim>
1561
+ </port>
1562
+ <port id="1" precision="FP32">
1563
+ <dim>1</dim>
1564
+ <dim>256</dim>
1565
+ <dim>1</dim>
1566
+ <dim>1</dim>
1567
+ </port>
1568
+ </input>
1569
+ <output>
1570
+ <port id="2" precision="FP32" names="/layer3/layer3.0/conv1/Conv_output_0">
1571
+ <dim>-1</dim>
1572
+ <dim>256</dim>
1573
+ <dim>14</dim>
1574
+ <dim>14</dim>
1575
+ </port>
1576
+ </output>
1577
+ </layer>
1578
+ <layer id="81" name="/layer3/layer3.0/relu/Relu" type="ReLU" version="opset1">
1579
+ <input>
1580
+ <port id="0" precision="FP32">
1581
+ <dim>-1</dim>
1582
+ <dim>256</dim>
1583
+ <dim>14</dim>
1584
+ <dim>14</dim>
1585
+ </port>
1586
+ </input>
1587
+ <output>
1588
+ <port id="1" precision="FP32" names="/layer3/layer3.0/relu/Relu_output_0">
1589
+ <dim>-1</dim>
1590
+ <dim>256</dim>
1591
+ <dim>14</dim>
1592
+ <dim>14</dim>
1593
+ </port>
1594
+ </output>
1595
+ </layer>
1596
+ <layer id="82" name="onnx::Conv_226_compressed" type="Const" version="opset1">
1597
+ <data element_type="f16" shape="256, 256, 3, 3" offset="1954560" size="1179648" />
1598
+ <output>
1599
+ <port id="0" precision="FP16" names="onnx::Conv_226">
1600
+ <dim>256</dim>
1601
+ <dim>256</dim>
1602
+ <dim>3</dim>
1603
+ <dim>3</dim>
1604
+ </port>
1605
+ </output>
1606
+ </layer>
1607
+ <layer id="83" name="onnx::Conv_226" type="Convert" version="opset1">
1608
+ <data destination_type="f32" />
1609
+ <rt_info>
1610
+ <attribute name="decompression" version="0" />
1611
+ </rt_info>
1612
+ <input>
1613
+ <port id="0" precision="FP16">
1614
+ <dim>256</dim>
1615
+ <dim>256</dim>
1616
+ <dim>3</dim>
1617
+ <dim>3</dim>
1618
+ </port>
1619
+ </input>
1620
+ <output>
1621
+ <port id="1" precision="FP32">
1622
+ <dim>256</dim>
1623
+ <dim>256</dim>
1624
+ <dim>3</dim>
1625
+ <dim>3</dim>
1626
+ </port>
1627
+ </output>
1628
+ </layer>
1629
+ <layer id="84" name="/layer3/layer3.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
1630
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1631
+ <input>
1632
+ <port id="0" precision="FP32">
1633
+ <dim>-1</dim>
1634
+ <dim>256</dim>
1635
+ <dim>14</dim>
1636
+ <dim>14</dim>
1637
+ </port>
1638
+ <port id="1" precision="FP32">
1639
+ <dim>256</dim>
1640
+ <dim>256</dim>
1641
+ <dim>3</dim>
1642
+ <dim>3</dim>
1643
+ </port>
1644
+ </input>
1645
+ <output>
1646
+ <port id="2" precision="FP32">
1647
+ <dim>-1</dim>
1648
+ <dim>256</dim>
1649
+ <dim>14</dim>
1650
+ <dim>14</dim>
1651
+ </port>
1652
+ </output>
1653
+ </layer>
1654
+ <layer id="85" name="Reshape_210_compressed" type="Const" version="opset1">
1655
+ <data element_type="f16" shape="1, 256, 1, 1" offset="3134208" size="512" />
1656
+ <output>
1657
+ <port id="0" precision="FP16">
1658
+ <dim>1</dim>
1659
+ <dim>256</dim>
1660
+ <dim>1</dim>
1661
+ <dim>1</dim>
1662
+ </port>
1663
+ </output>
1664
+ </layer>
1665
+ <layer id="86" name="Reshape_210" type="Convert" version="opset1">
1666
+ <data destination_type="f32" />
1667
+ <rt_info>
1668
+ <attribute name="decompression" version="0" />
1669
+ </rt_info>
1670
+ <input>
1671
+ <port id="0" precision="FP16">
1672
+ <dim>1</dim>
1673
+ <dim>256</dim>
1674
+ <dim>1</dim>
1675
+ <dim>1</dim>
1676
+ </port>
1677
+ </input>
1678
+ <output>
1679
+ <port id="1" precision="FP32">
1680
+ <dim>1</dim>
1681
+ <dim>256</dim>
1682
+ <dim>1</dim>
1683
+ <dim>1</dim>
1684
+ </port>
1685
+ </output>
1686
+ </layer>
1687
+ <layer id="87" name="/layer3/layer3.0/conv2/Conv" type="Add" version="opset1">
1688
+ <data auto_broadcast="numpy" />
1689
+ <input>
1690
+ <port id="0" precision="FP32">
1691
+ <dim>-1</dim>
1692
+ <dim>256</dim>
1693
+ <dim>14</dim>
1694
+ <dim>14</dim>
1695
+ </port>
1696
+ <port id="1" precision="FP32">
1697
+ <dim>1</dim>
1698
+ <dim>256</dim>
1699
+ <dim>1</dim>
1700
+ <dim>1</dim>
1701
+ </port>
1702
+ </input>
1703
+ <output>
1704
+ <port id="2" precision="FP32" names="/layer3/layer3.0/conv2/Conv_output_0">
1705
+ <dim>-1</dim>
1706
+ <dim>256</dim>
1707
+ <dim>14</dim>
1708
+ <dim>14</dim>
1709
+ </port>
1710
+ </output>
1711
+ </layer>
1712
+ <layer id="88" name="onnx::Conv_229_compressed" type="Const" version="opset1">
1713
+ <data element_type="f16" shape="256, 128, 1, 1" offset="3134720" size="65536" />
1714
+ <output>
1715
+ <port id="0" precision="FP16" names="onnx::Conv_229">
1716
+ <dim>256</dim>
1717
+ <dim>128</dim>
1718
+ <dim>1</dim>
1719
+ <dim>1</dim>
1720
+ </port>
1721
+ </output>
1722
+ </layer>
1723
+ <layer id="89" name="onnx::Conv_229" type="Convert" version="opset1">
1724
+ <data destination_type="f32" />
1725
+ <rt_info>
1726
+ <attribute name="decompression" version="0" />
1727
+ </rt_info>
1728
+ <input>
1729
+ <port id="0" precision="FP16">
1730
+ <dim>256</dim>
1731
+ <dim>128</dim>
1732
+ <dim>1</dim>
1733
+ <dim>1</dim>
1734
+ </port>
1735
+ </input>
1736
+ <output>
1737
+ <port id="1" precision="FP32">
1738
+ <dim>256</dim>
1739
+ <dim>128</dim>
1740
+ <dim>1</dim>
1741
+ <dim>1</dim>
1742
+ </port>
1743
+ </output>
1744
+ </layer>
1745
+ <layer id="90" name="/layer3/layer3.0/downsample/downsample.0/Conv/WithoutBiases" type="Convolution" version="opset1">
1746
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
1747
+ <input>
1748
+ <port id="0" precision="FP32">
1749
+ <dim>-1</dim>
1750
+ <dim>128</dim>
1751
+ <dim>28</dim>
1752
+ <dim>28</dim>
1753
+ </port>
1754
+ <port id="1" precision="FP32">
1755
+ <dim>256</dim>
1756
+ <dim>128</dim>
1757
+ <dim>1</dim>
1758
+ <dim>1</dim>
1759
+ </port>
1760
+ </input>
1761
+ <output>
1762
+ <port id="2" precision="FP32">
1763
+ <dim>-1</dim>
1764
+ <dim>256</dim>
1765
+ <dim>14</dim>
1766
+ <dim>14</dim>
1767
+ </port>
1768
+ </output>
1769
+ </layer>
1770
+ <layer id="91" name="Reshape_194_compressed" type="Const" version="opset1">
1771
+ <data element_type="f16" shape="1, 256, 1, 1" offset="3200256" size="512" />
1772
+ <output>
1773
+ <port id="0" precision="FP16">
1774
+ <dim>1</dim>
1775
+ <dim>256</dim>
1776
+ <dim>1</dim>
1777
+ <dim>1</dim>
1778
+ </port>
1779
+ </output>
1780
+ </layer>
1781
+ <layer id="92" name="Reshape_194" type="Convert" version="opset1">
1782
+ <data destination_type="f32" />
1783
+ <rt_info>
1784
+ <attribute name="decompression" version="0" />
1785
+ </rt_info>
1786
+ <input>
1787
+ <port id="0" precision="FP16">
1788
+ <dim>1</dim>
1789
+ <dim>256</dim>
1790
+ <dim>1</dim>
1791
+ <dim>1</dim>
1792
+ </port>
1793
+ </input>
1794
+ <output>
1795
+ <port id="1" precision="FP32">
1796
+ <dim>1</dim>
1797
+ <dim>256</dim>
1798
+ <dim>1</dim>
1799
+ <dim>1</dim>
1800
+ </port>
1801
+ </output>
1802
+ </layer>
1803
+ <layer id="93" name="/layer3/layer3.0/downsample/downsample.0/Conv" type="Add" version="opset1">
1804
+ <data auto_broadcast="numpy" />
1805
+ <input>
1806
+ <port id="0" precision="FP32">
1807
+ <dim>-1</dim>
1808
+ <dim>256</dim>
1809
+ <dim>14</dim>
1810
+ <dim>14</dim>
1811
+ </port>
1812
+ <port id="1" precision="FP32">
1813
+ <dim>1</dim>
1814
+ <dim>256</dim>
1815
+ <dim>1</dim>
1816
+ <dim>1</dim>
1817
+ </port>
1818
+ </input>
1819
+ <output>
1820
+ <port id="2" precision="FP32" names="/layer3/layer3.0/downsample/downsample.0/Conv_output_0">
1821
+ <dim>-1</dim>
1822
+ <dim>256</dim>
1823
+ <dim>14</dim>
1824
+ <dim>14</dim>
1825
+ </port>
1826
+ </output>
1827
+ </layer>
1828
+ <layer id="94" name="/layer3/layer3.0/Add" type="Add" version="opset1">
1829
+ <data auto_broadcast="numpy" />
1830
+ <input>
1831
+ <port id="0" precision="FP32">
1832
+ <dim>-1</dim>
1833
+ <dim>256</dim>
1834
+ <dim>14</dim>
1835
+ <dim>14</dim>
1836
+ </port>
1837
+ <port id="1" precision="FP32">
1838
+ <dim>-1</dim>
1839
+ <dim>256</dim>
1840
+ <dim>14</dim>
1841
+ <dim>14</dim>
1842
+ </port>
1843
+ </input>
1844
+ <output>
1845
+ <port id="2" precision="FP32" names="/layer3/layer3.0/Add_output_0">
1846
+ <dim>-1</dim>
1847
+ <dim>256</dim>
1848
+ <dim>14</dim>
1849
+ <dim>14</dim>
1850
+ </port>
1851
+ </output>
1852
+ </layer>
1853
+ <layer id="95" name="/layer3/layer3.0/relu_1/Relu" type="ReLU" version="opset1">
1854
+ <input>
1855
+ <port id="0" precision="FP32">
1856
+ <dim>-1</dim>
1857
+ <dim>256</dim>
1858
+ <dim>14</dim>
1859
+ <dim>14</dim>
1860
+ </port>
1861
+ </input>
1862
+ <output>
1863
+ <port id="1" precision="FP32" names="/layer3/layer3.0/relu_1/Relu_output_0">
1864
+ <dim>-1</dim>
1865
+ <dim>256</dim>
1866
+ <dim>14</dim>
1867
+ <dim>14</dim>
1868
+ </port>
1869
+ </output>
1870
+ </layer>
1871
+ <layer id="96" name="onnx::Conv_232_compressed" type="Const" version="opset1">
1872
+ <data element_type="f16" shape="256, 256, 3, 3" offset="3200768" size="1179648" />
1873
+ <output>
1874
+ <port id="0" precision="FP16" names="onnx::Conv_232">
1875
+ <dim>256</dim>
1876
+ <dim>256</dim>
1877
+ <dim>3</dim>
1878
+ <dim>3</dim>
1879
+ </port>
1880
+ </output>
1881
+ </layer>
1882
+ <layer id="97" name="onnx::Conv_232" type="Convert" version="opset1">
1883
+ <data destination_type="f32" />
1884
+ <rt_info>
1885
+ <attribute name="decompression" version="0" />
1886
+ </rt_info>
1887
+ <input>
1888
+ <port id="0" precision="FP16">
1889
+ <dim>256</dim>
1890
+ <dim>256</dim>
1891
+ <dim>3</dim>
1892
+ <dim>3</dim>
1893
+ </port>
1894
+ </input>
1895
+ <output>
1896
+ <port id="1" precision="FP32">
1897
+ <dim>256</dim>
1898
+ <dim>256</dim>
1899
+ <dim>3</dim>
1900
+ <dim>3</dim>
1901
+ </port>
1902
+ </output>
1903
+ </layer>
1904
+ <layer id="98" name="/layer3/layer3.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
1905
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
1906
+ <input>
1907
+ <port id="0" precision="FP32">
1908
+ <dim>-1</dim>
1909
+ <dim>256</dim>
1910
+ <dim>14</dim>
1911
+ <dim>14</dim>
1912
+ </port>
1913
+ <port id="1" precision="FP32">
1914
+ <dim>256</dim>
1915
+ <dim>256</dim>
1916
+ <dim>3</dim>
1917
+ <dim>3</dim>
1918
+ </port>
1919
+ </input>
1920
+ <output>
1921
+ <port id="2" precision="FP32">
1922
+ <dim>-1</dim>
1923
+ <dim>256</dim>
1924
+ <dim>14</dim>
1925
+ <dim>14</dim>
1926
+ </port>
1927
+ </output>
1928
+ </layer>
1929
+ <layer id="99" name="Reshape_227_compressed" type="Const" version="opset1">
1930
+ <data element_type="f16" shape="1, 256, 1, 1" offset="4380416" size="512" />
1931
+ <output>
1932
+ <port id="0" precision="FP16">
1933
+ <dim>1</dim>
1934
+ <dim>256</dim>
1935
+ <dim>1</dim>
1936
+ <dim>1</dim>
1937
+ </port>
1938
+ </output>
1939
+ </layer>
1940
+ <layer id="100" name="Reshape_227" type="Convert" version="opset1">
1941
+ <data destination_type="f32" />
1942
+ <rt_info>
1943
+ <attribute name="decompression" version="0" />
1944
+ </rt_info>
1945
+ <input>
1946
+ <port id="0" precision="FP16">
1947
+ <dim>1</dim>
1948
+ <dim>256</dim>
1949
+ <dim>1</dim>
1950
+ <dim>1</dim>
1951
+ </port>
1952
+ </input>
1953
+ <output>
1954
+ <port id="1" precision="FP32">
1955
+ <dim>1</dim>
1956
+ <dim>256</dim>
1957
+ <dim>1</dim>
1958
+ <dim>1</dim>
1959
+ </port>
1960
+ </output>
1961
+ </layer>
1962
+ <layer id="101" name="/layer3/layer3.1/conv1/Conv" type="Add" version="opset1">
1963
+ <data auto_broadcast="numpy" />
1964
+ <input>
1965
+ <port id="0" precision="FP32">
1966
+ <dim>-1</dim>
1967
+ <dim>256</dim>
1968
+ <dim>14</dim>
1969
+ <dim>14</dim>
1970
+ </port>
1971
+ <port id="1" precision="FP32">
1972
+ <dim>1</dim>
1973
+ <dim>256</dim>
1974
+ <dim>1</dim>
1975
+ <dim>1</dim>
1976
+ </port>
1977
+ </input>
1978
+ <output>
1979
+ <port id="2" precision="FP32" names="/layer3/layer3.1/conv1/Conv_output_0">
1980
+ <dim>-1</dim>
1981
+ <dim>256</dim>
1982
+ <dim>14</dim>
1983
+ <dim>14</dim>
1984
+ </port>
1985
+ </output>
1986
+ </layer>
1987
+ <layer id="102" name="/layer3/layer3.1/relu/Relu" type="ReLU" version="opset1">
1988
+ <input>
1989
+ <port id="0" precision="FP32">
1990
+ <dim>-1</dim>
1991
+ <dim>256</dim>
1992
+ <dim>14</dim>
1993
+ <dim>14</dim>
1994
+ </port>
1995
+ </input>
1996
+ <output>
1997
+ <port id="1" precision="FP32" names="/layer3/layer3.1/relu/Relu_output_0">
1998
+ <dim>-1</dim>
1999
+ <dim>256</dim>
2000
+ <dim>14</dim>
2001
+ <dim>14</dim>
2002
+ </port>
2003
+ </output>
2004
+ </layer>
2005
+ <layer id="103" name="onnx::Conv_235_compressed" type="Const" version="opset1">
2006
+ <data element_type="f16" shape="256, 256, 3, 3" offset="4380928" size="1179648" />
2007
+ <output>
2008
+ <port id="0" precision="FP16" names="onnx::Conv_235">
2009
+ <dim>256</dim>
2010
+ <dim>256</dim>
2011
+ <dim>3</dim>
2012
+ <dim>3</dim>
2013
+ </port>
2014
+ </output>
2015
+ </layer>
2016
+ <layer id="104" name="onnx::Conv_235" type="Convert" version="opset1">
2017
+ <data destination_type="f32" />
2018
+ <rt_info>
2019
+ <attribute name="decompression" version="0" />
2020
+ </rt_info>
2021
+ <input>
2022
+ <port id="0" precision="FP16">
2023
+ <dim>256</dim>
2024
+ <dim>256</dim>
2025
+ <dim>3</dim>
2026
+ <dim>3</dim>
2027
+ </port>
2028
+ </input>
2029
+ <output>
2030
+ <port id="1" precision="FP32">
2031
+ <dim>256</dim>
2032
+ <dim>256</dim>
2033
+ <dim>3</dim>
2034
+ <dim>3</dim>
2035
+ </port>
2036
+ </output>
2037
+ </layer>
2038
+ <layer id="105" name="/layer3/layer3.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
2039
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
2040
+ <input>
2041
+ <port id="0" precision="FP32">
2042
+ <dim>-1</dim>
2043
+ <dim>256</dim>
2044
+ <dim>14</dim>
2045
+ <dim>14</dim>
2046
+ </port>
2047
+ <port id="1" precision="FP32">
2048
+ <dim>256</dim>
2049
+ <dim>256</dim>
2050
+ <dim>3</dim>
2051
+ <dim>3</dim>
2052
+ </port>
2053
+ </input>
2054
+ <output>
2055
+ <port id="2" precision="FP32">
2056
+ <dim>-1</dim>
2057
+ <dim>256</dim>
2058
+ <dim>14</dim>
2059
+ <dim>14</dim>
2060
+ </port>
2061
+ </output>
2062
+ </layer>
2063
+ <layer id="106" name="Reshape_243_compressed" type="Const" version="opset1">
2064
+ <data element_type="f16" shape="1, 256, 1, 1" offset="5560576" size="512" />
2065
+ <output>
2066
+ <port id="0" precision="FP16">
2067
+ <dim>1</dim>
2068
+ <dim>256</dim>
2069
+ <dim>1</dim>
2070
+ <dim>1</dim>
2071
+ </port>
2072
+ </output>
2073
+ </layer>
2074
+ <layer id="107" name="Reshape_243" type="Convert" version="opset1">
2075
+ <data destination_type="f32" />
2076
+ <rt_info>
2077
+ <attribute name="decompression" version="0" />
2078
+ </rt_info>
2079
+ <input>
2080
+ <port id="0" precision="FP16">
2081
+ <dim>1</dim>
2082
+ <dim>256</dim>
2083
+ <dim>1</dim>
2084
+ <dim>1</dim>
2085
+ </port>
2086
+ </input>
2087
+ <output>
2088
+ <port id="1" precision="FP32">
2089
+ <dim>1</dim>
2090
+ <dim>256</dim>
2091
+ <dim>1</dim>
2092
+ <dim>1</dim>
2093
+ </port>
2094
+ </output>
2095
+ </layer>
2096
+ <layer id="108" name="/layer3/layer3.1/conv2/Conv" type="Add" version="opset1">
2097
+ <data auto_broadcast="numpy" />
2098
+ <input>
2099
+ <port id="0" precision="FP32">
2100
+ <dim>-1</dim>
2101
+ <dim>256</dim>
2102
+ <dim>14</dim>
2103
+ <dim>14</dim>
2104
+ </port>
2105
+ <port id="1" precision="FP32">
2106
+ <dim>1</dim>
2107
+ <dim>256</dim>
2108
+ <dim>1</dim>
2109
+ <dim>1</dim>
2110
+ </port>
2111
+ </input>
2112
+ <output>
2113
+ <port id="2" precision="FP32" names="/layer3/layer3.1/conv2/Conv_output_0">
2114
+ <dim>-1</dim>
2115
+ <dim>256</dim>
2116
+ <dim>14</dim>
2117
+ <dim>14</dim>
2118
+ </port>
2119
+ </output>
2120
+ </layer>
2121
+ <layer id="109" name="/layer3/layer3.1/Add" type="Add" version="opset1">
2122
+ <data auto_broadcast="numpy" />
2123
+ <input>
2124
+ <port id="0" precision="FP32">
2125
+ <dim>-1</dim>
2126
+ <dim>256</dim>
2127
+ <dim>14</dim>
2128
+ <dim>14</dim>
2129
+ </port>
2130
+ <port id="1" precision="FP32">
2131
+ <dim>-1</dim>
2132
+ <dim>256</dim>
2133
+ <dim>14</dim>
2134
+ <dim>14</dim>
2135
+ </port>
2136
+ </input>
2137
+ <output>
2138
+ <port id="2" precision="FP32" names="/layer3/layer3.1/Add_output_0">
2139
+ <dim>-1</dim>
2140
+ <dim>256</dim>
2141
+ <dim>14</dim>
2142
+ <dim>14</dim>
2143
+ </port>
2144
+ </output>
2145
+ </layer>
2146
+ <layer id="110" name="/layer3/layer3.1/relu_1/Relu" type="ReLU" version="opset1">
2147
+ <input>
2148
+ <port id="0" precision="FP32">
2149
+ <dim>-1</dim>
2150
+ <dim>256</dim>
2151
+ <dim>14</dim>
2152
+ <dim>14</dim>
2153
+ </port>
2154
+ </input>
2155
+ <output>
2156
+ <port id="1" precision="FP32" names="/layer3/layer3.1/relu_1/Relu_output_0">
2157
+ <dim>-1</dim>
2158
+ <dim>256</dim>
2159
+ <dim>14</dim>
2160
+ <dim>14</dim>
2161
+ </port>
2162
+ </output>
2163
+ </layer>
2164
+ <layer id="111" name="onnx::Conv_238_compressed" type="Const" version="opset1">
2165
+ <data element_type="f16" shape="512, 256, 3, 3" offset="5561088" size="2359296" />
2166
+ <output>
2167
+ <port id="0" precision="FP16" names="onnx::Conv_238">
2168
+ <dim>512</dim>
2169
+ <dim>256</dim>
2170
+ <dim>3</dim>
2171
+ <dim>3</dim>
2172
+ </port>
2173
+ </output>
2174
+ </layer>
2175
+ <layer id="112" name="onnx::Conv_238" type="Convert" version="opset1">
2176
+ <data destination_type="f32" />
2177
+ <rt_info>
2178
+ <attribute name="decompression" version="0" />
2179
+ </rt_info>
2180
+ <input>
2181
+ <port id="0" precision="FP16">
2182
+ <dim>512</dim>
2183
+ <dim>256</dim>
2184
+ <dim>3</dim>
2185
+ <dim>3</dim>
2186
+ </port>
2187
+ </input>
2188
+ <output>
2189
+ <port id="1" precision="FP32">
2190
+ <dim>512</dim>
2191
+ <dim>256</dim>
2192
+ <dim>3</dim>
2193
+ <dim>3</dim>
2194
+ </port>
2195
+ </output>
2196
+ </layer>
2197
+ <layer id="113" name="/layer4/layer4.0/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
2198
+ <data strides="2, 2" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
2199
+ <input>
2200
+ <port id="0" precision="FP32">
2201
+ <dim>-1</dim>
2202
+ <dim>256</dim>
2203
+ <dim>14</dim>
2204
+ <dim>14</dim>
2205
+ </port>
2206
+ <port id="1" precision="FP32">
2207
+ <dim>512</dim>
2208
+ <dim>256</dim>
2209
+ <dim>3</dim>
2210
+ <dim>3</dim>
2211
+ </port>
2212
+ </input>
2213
+ <output>
2214
+ <port id="2" precision="FP32">
2215
+ <dim>-1</dim>
2216
+ <dim>512</dim>
2217
+ <dim>7</dim>
2218
+ <dim>7</dim>
2219
+ </port>
2220
+ </output>
2221
+ </layer>
2222
+ <layer id="114" name="Reshape_260_compressed" type="Const" version="opset1">
2223
+ <data element_type="f16" shape="1, 512, 1, 1" offset="7920384" size="1024" />
2224
+ <output>
2225
+ <port id="0" precision="FP16">
2226
+ <dim>1</dim>
2227
+ <dim>512</dim>
2228
+ <dim>1</dim>
2229
+ <dim>1</dim>
2230
+ </port>
2231
+ </output>
2232
+ </layer>
2233
+ <layer id="115" name="Reshape_260" type="Convert" version="opset1">
2234
+ <data destination_type="f32" />
2235
+ <rt_info>
2236
+ <attribute name="decompression" version="0" />
2237
+ </rt_info>
2238
+ <input>
2239
+ <port id="0" precision="FP16">
2240
+ <dim>1</dim>
2241
+ <dim>512</dim>
2242
+ <dim>1</dim>
2243
+ <dim>1</dim>
2244
+ </port>
2245
+ </input>
2246
+ <output>
2247
+ <port id="1" precision="FP32">
2248
+ <dim>1</dim>
2249
+ <dim>512</dim>
2250
+ <dim>1</dim>
2251
+ <dim>1</dim>
2252
+ </port>
2253
+ </output>
2254
+ </layer>
2255
+ <layer id="116" name="/layer4/layer4.0/conv1/Conv" type="Add" version="opset1">
2256
+ <data auto_broadcast="numpy" />
2257
+ <input>
2258
+ <port id="0" precision="FP32">
2259
+ <dim>-1</dim>
2260
+ <dim>512</dim>
2261
+ <dim>7</dim>
2262
+ <dim>7</dim>
2263
+ </port>
2264
+ <port id="1" precision="FP32">
2265
+ <dim>1</dim>
2266
+ <dim>512</dim>
2267
+ <dim>1</dim>
2268
+ <dim>1</dim>
2269
+ </port>
2270
+ </input>
2271
+ <output>
2272
+ <port id="2" precision="FP32" names="/layer4/layer4.0/conv1/Conv_output_0">
2273
+ <dim>-1</dim>
2274
+ <dim>512</dim>
2275
+ <dim>7</dim>
2276
+ <dim>7</dim>
2277
+ </port>
2278
+ </output>
2279
+ </layer>
2280
+ <layer id="117" name="/layer4/layer4.0/relu/Relu" type="ReLU" version="opset1">
2281
+ <input>
2282
+ <port id="0" precision="FP32">
2283
+ <dim>-1</dim>
2284
+ <dim>512</dim>
2285
+ <dim>7</dim>
2286
+ <dim>7</dim>
2287
+ </port>
2288
+ </input>
2289
+ <output>
2290
+ <port id="1" precision="FP32" names="/layer4/layer4.0/relu/Relu_output_0">
2291
+ <dim>-1</dim>
2292
+ <dim>512</dim>
2293
+ <dim>7</dim>
2294
+ <dim>7</dim>
2295
+ </port>
2296
+ </output>
2297
+ </layer>
2298
+ <layer id="118" name="onnx::Conv_241_compressed" type="Const" version="opset1">
2299
+ <data element_type="f16" shape="512, 512, 3, 3" offset="7921408" size="4718592" />
2300
+ <output>
2301
+ <port id="0" precision="FP16" names="onnx::Conv_241">
2302
+ <dim>512</dim>
2303
+ <dim>512</dim>
2304
+ <dim>3</dim>
2305
+ <dim>3</dim>
2306
+ </port>
2307
+ </output>
2308
+ </layer>
2309
+ <layer id="119" name="onnx::Conv_241" type="Convert" version="opset1">
2310
+ <data destination_type="f32" />
2311
+ <rt_info>
2312
+ <attribute name="decompression" version="0" />
2313
+ </rt_info>
2314
+ <input>
2315
+ <port id="0" precision="FP16">
2316
+ <dim>512</dim>
2317
+ <dim>512</dim>
2318
+ <dim>3</dim>
2319
+ <dim>3</dim>
2320
+ </port>
2321
+ </input>
2322
+ <output>
2323
+ <port id="1" precision="FP32">
2324
+ <dim>512</dim>
2325
+ <dim>512</dim>
2326
+ <dim>3</dim>
2327
+ <dim>3</dim>
2328
+ </port>
2329
+ </output>
2330
+ </layer>
2331
+ <layer id="120" name="/layer4/layer4.0/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
2332
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
2333
+ <input>
2334
+ <port id="0" precision="FP32">
2335
+ <dim>-1</dim>
2336
+ <dim>512</dim>
2337
+ <dim>7</dim>
2338
+ <dim>7</dim>
2339
+ </port>
2340
+ <port id="1" precision="FP32">
2341
+ <dim>512</dim>
2342
+ <dim>512</dim>
2343
+ <dim>3</dim>
2344
+ <dim>3</dim>
2345
+ </port>
2346
+ </input>
2347
+ <output>
2348
+ <port id="2" precision="FP32">
2349
+ <dim>-1</dim>
2350
+ <dim>512</dim>
2351
+ <dim>7</dim>
2352
+ <dim>7</dim>
2353
+ </port>
2354
+ </output>
2355
+ </layer>
2356
+ <layer id="121" name="Reshape_291_compressed" type="Const" version="opset1">
2357
+ <data element_type="f16" shape="1, 512, 1, 1" offset="12640000" size="1024" />
2358
+ <output>
2359
+ <port id="0" precision="FP16">
2360
+ <dim>1</dim>
2361
+ <dim>512</dim>
2362
+ <dim>1</dim>
2363
+ <dim>1</dim>
2364
+ </port>
2365
+ </output>
2366
+ </layer>
2367
+ <layer id="122" name="Reshape_291" type="Convert" version="opset1">
2368
+ <data destination_type="f32" />
2369
+ <rt_info>
2370
+ <attribute name="decompression" version="0" />
2371
+ </rt_info>
2372
+ <input>
2373
+ <port id="0" precision="FP16">
2374
+ <dim>1</dim>
2375
+ <dim>512</dim>
2376
+ <dim>1</dim>
2377
+ <dim>1</dim>
2378
+ </port>
2379
+ </input>
2380
+ <output>
2381
+ <port id="1" precision="FP32">
2382
+ <dim>1</dim>
2383
+ <dim>512</dim>
2384
+ <dim>1</dim>
2385
+ <dim>1</dim>
2386
+ </port>
2387
+ </output>
2388
+ </layer>
2389
+ <layer id="123" name="/layer4/layer4.0/conv2/Conv" type="Add" version="opset1">
2390
+ <data auto_broadcast="numpy" />
2391
+ <input>
2392
+ <port id="0" precision="FP32">
2393
+ <dim>-1</dim>
2394
+ <dim>512</dim>
2395
+ <dim>7</dim>
2396
+ <dim>7</dim>
2397
+ </port>
2398
+ <port id="1" precision="FP32">
2399
+ <dim>1</dim>
2400
+ <dim>512</dim>
2401
+ <dim>1</dim>
2402
+ <dim>1</dim>
2403
+ </port>
2404
+ </input>
2405
+ <output>
2406
+ <port id="2" precision="FP32" names="/layer4/layer4.0/conv2/Conv_output_0">
2407
+ <dim>-1</dim>
2408
+ <dim>512</dim>
2409
+ <dim>7</dim>
2410
+ <dim>7</dim>
2411
+ </port>
2412
+ </output>
2413
+ </layer>
2414
+ <layer id="124" name="onnx::Conv_244_compressed" type="Const" version="opset1">
2415
+ <data element_type="f16" shape="512, 256, 1, 1" offset="12641024" size="262144" />
2416
+ <output>
2417
+ <port id="0" precision="FP16" names="onnx::Conv_244">
2418
+ <dim>512</dim>
2419
+ <dim>256</dim>
2420
+ <dim>1</dim>
2421
+ <dim>1</dim>
2422
+ </port>
2423
+ </output>
2424
+ </layer>
2425
+ <layer id="125" name="onnx::Conv_244" type="Convert" version="opset1">
2426
+ <data destination_type="f32" />
2427
+ <rt_info>
2428
+ <attribute name="decompression" version="0" />
2429
+ </rt_info>
2430
+ <input>
2431
+ <port id="0" precision="FP16">
2432
+ <dim>512</dim>
2433
+ <dim>256</dim>
2434
+ <dim>1</dim>
2435
+ <dim>1</dim>
2436
+ </port>
2437
+ </input>
2438
+ <output>
2439
+ <port id="1" precision="FP32">
2440
+ <dim>512</dim>
2441
+ <dim>256</dim>
2442
+ <dim>1</dim>
2443
+ <dim>1</dim>
2444
+ </port>
2445
+ </output>
2446
+ </layer>
2447
+ <layer id="126" name="/layer4/layer4.0/downsample/downsample.0/Conv/WithoutBiases" type="Convolution" version="opset1">
2448
+ <data strides="2, 2" dilations="1, 1" pads_begin="0, 0" pads_end="0, 0" auto_pad="explicit" />
2449
+ <input>
2450
+ <port id="0" precision="FP32">
2451
+ <dim>-1</dim>
2452
+ <dim>256</dim>
2453
+ <dim>14</dim>
2454
+ <dim>14</dim>
2455
+ </port>
2456
+ <port id="1" precision="FP32">
2457
+ <dim>512</dim>
2458
+ <dim>256</dim>
2459
+ <dim>1</dim>
2460
+ <dim>1</dim>
2461
+ </port>
2462
+ </input>
2463
+ <output>
2464
+ <port id="2" precision="FP32">
2465
+ <dim>-1</dim>
2466
+ <dim>512</dim>
2467
+ <dim>7</dim>
2468
+ <dim>7</dim>
2469
+ </port>
2470
+ </output>
2471
+ </layer>
2472
+ <layer id="127" name="Reshape_275_compressed" type="Const" version="opset1">
2473
+ <data element_type="f16" shape="1, 512, 1, 1" offset="12903168" size="1024" />
2474
+ <output>
2475
+ <port id="0" precision="FP16">
2476
+ <dim>1</dim>
2477
+ <dim>512</dim>
2478
+ <dim>1</dim>
2479
+ <dim>1</dim>
2480
+ </port>
2481
+ </output>
2482
+ </layer>
2483
+ <layer id="128" name="Reshape_275" type="Convert" version="opset1">
2484
+ <data destination_type="f32" />
2485
+ <rt_info>
2486
+ <attribute name="decompression" version="0" />
2487
+ </rt_info>
2488
+ <input>
2489
+ <port id="0" precision="FP16">
2490
+ <dim>1</dim>
2491
+ <dim>512</dim>
2492
+ <dim>1</dim>
2493
+ <dim>1</dim>
2494
+ </port>
2495
+ </input>
2496
+ <output>
2497
+ <port id="1" precision="FP32">
2498
+ <dim>1</dim>
2499
+ <dim>512</dim>
2500
+ <dim>1</dim>
2501
+ <dim>1</dim>
2502
+ </port>
2503
+ </output>
2504
+ </layer>
2505
+ <layer id="129" name="/layer4/layer4.0/downsample/downsample.0/Conv" type="Add" version="opset1">
2506
+ <data auto_broadcast="numpy" />
2507
+ <input>
2508
+ <port id="0" precision="FP32">
2509
+ <dim>-1</dim>
2510
+ <dim>512</dim>
2511
+ <dim>7</dim>
2512
+ <dim>7</dim>
2513
+ </port>
2514
+ <port id="1" precision="FP32">
2515
+ <dim>1</dim>
2516
+ <dim>512</dim>
2517
+ <dim>1</dim>
2518
+ <dim>1</dim>
2519
+ </port>
2520
+ </input>
2521
+ <output>
2522
+ <port id="2" precision="FP32" names="/layer4/layer4.0/downsample/downsample.0/Conv_output_0">
2523
+ <dim>-1</dim>
2524
+ <dim>512</dim>
2525
+ <dim>7</dim>
2526
+ <dim>7</dim>
2527
+ </port>
2528
+ </output>
2529
+ </layer>
2530
+ <layer id="130" name="/layer4/layer4.0/Add" type="Add" version="opset1">
2531
+ <data auto_broadcast="numpy" />
2532
+ <input>
2533
+ <port id="0" precision="FP32">
2534
+ <dim>-1</dim>
2535
+ <dim>512</dim>
2536
+ <dim>7</dim>
2537
+ <dim>7</dim>
2538
+ </port>
2539
+ <port id="1" precision="FP32">
2540
+ <dim>-1</dim>
2541
+ <dim>512</dim>
2542
+ <dim>7</dim>
2543
+ <dim>7</dim>
2544
+ </port>
2545
+ </input>
2546
+ <output>
2547
+ <port id="2" precision="FP32" names="/layer4/layer4.0/Add_output_0">
2548
+ <dim>-1</dim>
2549
+ <dim>512</dim>
2550
+ <dim>7</dim>
2551
+ <dim>7</dim>
2552
+ </port>
2553
+ </output>
2554
+ </layer>
2555
+ <layer id="131" name="/layer4/layer4.0/relu_1/Relu" type="ReLU" version="opset1">
2556
+ <input>
2557
+ <port id="0" precision="FP32">
2558
+ <dim>-1</dim>
2559
+ <dim>512</dim>
2560
+ <dim>7</dim>
2561
+ <dim>7</dim>
2562
+ </port>
2563
+ </input>
2564
+ <output>
2565
+ <port id="1" precision="FP32" names="/layer4/layer4.0/relu_1/Relu_output_0">
2566
+ <dim>-1</dim>
2567
+ <dim>512</dim>
2568
+ <dim>7</dim>
2569
+ <dim>7</dim>
2570
+ </port>
2571
+ </output>
2572
+ </layer>
2573
+ <layer id="132" name="onnx::Conv_247_compressed" type="Const" version="opset1">
2574
+ <data element_type="f16" shape="512, 512, 3, 3" offset="12904192" size="4718592" />
2575
+ <output>
2576
+ <port id="0" precision="FP16" names="onnx::Conv_247">
2577
+ <dim>512</dim>
2578
+ <dim>512</dim>
2579
+ <dim>3</dim>
2580
+ <dim>3</dim>
2581
+ </port>
2582
+ </output>
2583
+ </layer>
2584
+ <layer id="133" name="onnx::Conv_247" type="Convert" version="opset1">
2585
+ <data destination_type="f32" />
2586
+ <rt_info>
2587
+ <attribute name="decompression" version="0" />
2588
+ </rt_info>
2589
+ <input>
2590
+ <port id="0" precision="FP16">
2591
+ <dim>512</dim>
2592
+ <dim>512</dim>
2593
+ <dim>3</dim>
2594
+ <dim>3</dim>
2595
+ </port>
2596
+ </input>
2597
+ <output>
2598
+ <port id="1" precision="FP32">
2599
+ <dim>512</dim>
2600
+ <dim>512</dim>
2601
+ <dim>3</dim>
2602
+ <dim>3</dim>
2603
+ </port>
2604
+ </output>
2605
+ </layer>
2606
+ <layer id="134" name="/layer4/layer4.1/conv1/Conv/WithoutBiases" type="Convolution" version="opset1">
2607
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
2608
+ <input>
2609
+ <port id="0" precision="FP32">
2610
+ <dim>-1</dim>
2611
+ <dim>512</dim>
2612
+ <dim>7</dim>
2613
+ <dim>7</dim>
2614
+ </port>
2615
+ <port id="1" precision="FP32">
2616
+ <dim>512</dim>
2617
+ <dim>512</dim>
2618
+ <dim>3</dim>
2619
+ <dim>3</dim>
2620
+ </port>
2621
+ </input>
2622
+ <output>
2623
+ <port id="2" precision="FP32">
2624
+ <dim>-1</dim>
2625
+ <dim>512</dim>
2626
+ <dim>7</dim>
2627
+ <dim>7</dim>
2628
+ </port>
2629
+ </output>
2630
+ </layer>
2631
+ <layer id="135" name="Reshape_308_compressed" type="Const" version="opset1">
2632
+ <data element_type="f16" shape="1, 512, 1, 1" offset="17622784" size="1024" />
2633
+ <output>
2634
+ <port id="0" precision="FP16">
2635
+ <dim>1</dim>
2636
+ <dim>512</dim>
2637
+ <dim>1</dim>
2638
+ <dim>1</dim>
2639
+ </port>
2640
+ </output>
2641
+ </layer>
2642
+ <layer id="136" name="Reshape_308" type="Convert" version="opset1">
2643
+ <data destination_type="f32" />
2644
+ <rt_info>
2645
+ <attribute name="decompression" version="0" />
2646
+ </rt_info>
2647
+ <input>
2648
+ <port id="0" precision="FP16">
2649
+ <dim>1</dim>
2650
+ <dim>512</dim>
2651
+ <dim>1</dim>
2652
+ <dim>1</dim>
2653
+ </port>
2654
+ </input>
2655
+ <output>
2656
+ <port id="1" precision="FP32">
2657
+ <dim>1</dim>
2658
+ <dim>512</dim>
2659
+ <dim>1</dim>
2660
+ <dim>1</dim>
2661
+ </port>
2662
+ </output>
2663
+ </layer>
2664
+ <layer id="137" name="/layer4/layer4.1/conv1/Conv" type="Add" version="opset1">
2665
+ <data auto_broadcast="numpy" />
2666
+ <input>
2667
+ <port id="0" precision="FP32">
2668
+ <dim>-1</dim>
2669
+ <dim>512</dim>
2670
+ <dim>7</dim>
2671
+ <dim>7</dim>
2672
+ </port>
2673
+ <port id="1" precision="FP32">
2674
+ <dim>1</dim>
2675
+ <dim>512</dim>
2676
+ <dim>1</dim>
2677
+ <dim>1</dim>
2678
+ </port>
2679
+ </input>
2680
+ <output>
2681
+ <port id="2" precision="FP32" names="/layer4/layer4.1/conv1/Conv_output_0">
2682
+ <dim>-1</dim>
2683
+ <dim>512</dim>
2684
+ <dim>7</dim>
2685
+ <dim>7</dim>
2686
+ </port>
2687
+ </output>
2688
+ </layer>
2689
+ <layer id="138" name="/layer4/layer4.1/relu/Relu" type="ReLU" version="opset1">
2690
+ <input>
2691
+ <port id="0" precision="FP32">
2692
+ <dim>-1</dim>
2693
+ <dim>512</dim>
2694
+ <dim>7</dim>
2695
+ <dim>7</dim>
2696
+ </port>
2697
+ </input>
2698
+ <output>
2699
+ <port id="1" precision="FP32" names="/layer4/layer4.1/relu/Relu_output_0">
2700
+ <dim>-1</dim>
2701
+ <dim>512</dim>
2702
+ <dim>7</dim>
2703
+ <dim>7</dim>
2704
+ </port>
2705
+ </output>
2706
+ </layer>
2707
+ <layer id="139" name="onnx::Conv_250_compressed" type="Const" version="opset1">
2708
+ <data element_type="f16" shape="512, 512, 3, 3" offset="17623808" size="4718592" />
2709
+ <output>
2710
+ <port id="0" precision="FP16" names="onnx::Conv_250">
2711
+ <dim>512</dim>
2712
+ <dim>512</dim>
2713
+ <dim>3</dim>
2714
+ <dim>3</dim>
2715
+ </port>
2716
+ </output>
2717
+ </layer>
2718
+ <layer id="140" name="onnx::Conv_250" type="Convert" version="opset1">
2719
+ <data destination_type="f32" />
2720
+ <rt_info>
2721
+ <attribute name="decompression" version="0" />
2722
+ </rt_info>
2723
+ <input>
2724
+ <port id="0" precision="FP16">
2725
+ <dim>512</dim>
2726
+ <dim>512</dim>
2727
+ <dim>3</dim>
2728
+ <dim>3</dim>
2729
+ </port>
2730
+ </input>
2731
+ <output>
2732
+ <port id="1" precision="FP32">
2733
+ <dim>512</dim>
2734
+ <dim>512</dim>
2735
+ <dim>3</dim>
2736
+ <dim>3</dim>
2737
+ </port>
2738
+ </output>
2739
+ </layer>
2740
+ <layer id="141" name="/layer4/layer4.1/conv2/Conv/WithoutBiases" type="Convolution" version="opset1">
2741
+ <data strides="1, 1" dilations="1, 1" pads_begin="1, 1" pads_end="1, 1" auto_pad="explicit" />
2742
+ <input>
2743
+ <port id="0" precision="FP32">
2744
+ <dim>-1</dim>
2745
+ <dim>512</dim>
2746
+ <dim>7</dim>
2747
+ <dim>7</dim>
2748
+ </port>
2749
+ <port id="1" precision="FP32">
2750
+ <dim>512</dim>
2751
+ <dim>512</dim>
2752
+ <dim>3</dim>
2753
+ <dim>3</dim>
2754
+ </port>
2755
+ </input>
2756
+ <output>
2757
+ <port id="2" precision="FP32">
2758
+ <dim>-1</dim>
2759
+ <dim>512</dim>
2760
+ <dim>7</dim>
2761
+ <dim>7</dim>
2762
+ </port>
2763
+ </output>
2764
+ </layer>
2765
+ <layer id="142" name="Reshape_324_compressed" type="Const" version="opset1">
2766
+ <data element_type="f16" shape="1, 512, 1, 1" offset="22342400" size="1024" />
2767
+ <output>
2768
+ <port id="0" precision="FP16">
2769
+ <dim>1</dim>
2770
+ <dim>512</dim>
2771
+ <dim>1</dim>
2772
+ <dim>1</dim>
2773
+ </port>
2774
+ </output>
2775
+ </layer>
2776
+ <layer id="143" name="Reshape_324" type="Convert" version="opset1">
2777
+ <data destination_type="f32" />
2778
+ <rt_info>
2779
+ <attribute name="decompression" version="0" />
2780
+ </rt_info>
2781
+ <input>
2782
+ <port id="0" precision="FP16">
2783
+ <dim>1</dim>
2784
+ <dim>512</dim>
2785
+ <dim>1</dim>
2786
+ <dim>1</dim>
2787
+ </port>
2788
+ </input>
2789
+ <output>
2790
+ <port id="1" precision="FP32">
2791
+ <dim>1</dim>
2792
+ <dim>512</dim>
2793
+ <dim>1</dim>
2794
+ <dim>1</dim>
2795
+ </port>
2796
+ </output>
2797
+ </layer>
2798
+ <layer id="144" name="/layer4/layer4.1/conv2/Conv" type="Add" version="opset1">
2799
+ <data auto_broadcast="numpy" />
2800
+ <input>
2801
+ <port id="0" precision="FP32">
2802
+ <dim>-1</dim>
2803
+ <dim>512</dim>
2804
+ <dim>7</dim>
2805
+ <dim>7</dim>
2806
+ </port>
2807
+ <port id="1" precision="FP32">
2808
+ <dim>1</dim>
2809
+ <dim>512</dim>
2810
+ <dim>1</dim>
2811
+ <dim>1</dim>
2812
+ </port>
2813
+ </input>
2814
+ <output>
2815
+ <port id="2" precision="FP32" names="/layer4/layer4.1/conv2/Conv_output_0">
2816
+ <dim>-1</dim>
2817
+ <dim>512</dim>
2818
+ <dim>7</dim>
2819
+ <dim>7</dim>
2820
+ </port>
2821
+ </output>
2822
+ </layer>
2823
+ <layer id="145" name="/layer4/layer4.1/Add" type="Add" version="opset1">
2824
+ <data auto_broadcast="numpy" />
2825
+ <input>
2826
+ <port id="0" precision="FP32">
2827
+ <dim>-1</dim>
2828
+ <dim>512</dim>
2829
+ <dim>7</dim>
2830
+ <dim>7</dim>
2831
+ </port>
2832
+ <port id="1" precision="FP32">
2833
+ <dim>-1</dim>
2834
+ <dim>512</dim>
2835
+ <dim>7</dim>
2836
+ <dim>7</dim>
2837
+ </port>
2838
+ </input>
2839
+ <output>
2840
+ <port id="2" precision="FP32" names="/layer4/layer4.1/Add_output_0">
2841
+ <dim>-1</dim>
2842
+ <dim>512</dim>
2843
+ <dim>7</dim>
2844
+ <dim>7</dim>
2845
+ </port>
2846
+ </output>
2847
+ </layer>
2848
+ <layer id="146" name="/layer4/layer4.1/relu_1/Relu" type="ReLU" version="opset1">
2849
+ <input>
2850
+ <port id="0" precision="FP32">
2851
+ <dim>-1</dim>
2852
+ <dim>512</dim>
2853
+ <dim>7</dim>
2854
+ <dim>7</dim>
2855
+ </port>
2856
+ </input>
2857
+ <output>
2858
+ <port id="1" precision="FP32" names="/layer4/layer4.1/relu_1/Relu_output_0">
2859
+ <dim>-1</dim>
2860
+ <dim>512</dim>
2861
+ <dim>7</dim>
2862
+ <dim>7</dim>
2863
+ </port>
2864
+ </output>
2865
+ </layer>
2866
+ <layer id="147" name="Range_334" type="Const" version="opset1">
2867
+ <data element_type="i64" shape="2" offset="22343424" size="16" />
2868
+ <output>
2869
+ <port id="0" precision="I64">
2870
+ <dim>2</dim>
2871
+ </port>
2872
+ </output>
2873
+ </layer>
2874
+ <layer id="148" name="/avgpool/GlobalAveragePool" type="ReduceMean" version="opset1">
2875
+ <data keep_dims="true" />
2876
+ <input>
2877
+ <port id="0" precision="FP32">
2878
+ <dim>-1</dim>
2879
+ <dim>512</dim>
2880
+ <dim>7</dim>
2881
+ <dim>7</dim>
2882
+ </port>
2883
+ <port id="1" precision="I64">
2884
+ <dim>2</dim>
2885
+ </port>
2886
+ </input>
2887
+ <output>
2888
+ <port id="2" precision="FP32" names="/avgpool/GlobalAveragePool_output_0">
2889
+ <dim>-1</dim>
2890
+ <dim>512</dim>
2891
+ <dim>1</dim>
2892
+ <dim>1</dim>
2893
+ </port>
2894
+ </output>
2895
+ </layer>
2896
+ <layer id="149" name="Concat_751" type="Const" version="opset1">
2897
+ <data element_type="i64" shape="2" offset="22343440" size="16" />
2898
+ <rt_info>
2899
+ <attribute name="precise" version="0" />
2900
+ </rt_info>
2901
+ <output>
2902
+ <port id="0" precision="I64">
2903
+ <dim>2</dim>
2904
+ </port>
2905
+ </output>
2906
+ </layer>
2907
+ <layer id="150" name="/Flatten" type="Reshape" version="opset1">
2908
+ <data special_zero="true" />
2909
+ <input>
2910
+ <port id="0" precision="FP32">
2911
+ <dim>-1</dim>
2912
+ <dim>512</dim>
2913
+ <dim>1</dim>
2914
+ <dim>1</dim>
2915
+ </port>
2916
+ <port id="1" precision="I64">
2917
+ <dim>2</dim>
2918
+ </port>
2919
+ </input>
2920
+ <output>
2921
+ <port id="2" precision="FP32" names="/Flatten_output_0">
2922
+ <dim>-1</dim>
2923
+ <dim>512</dim>
2924
+ </port>
2925
+ </output>
2926
+ </layer>
2927
+ <layer id="151" name="fc.weight_compressed" type="Const" version="opset1">
2928
+ <data element_type="f16" shape="10, 512" offset="22343456" size="10240" />
2929
+ <output>
2930
+ <port id="0" precision="FP16" names="fc.weight">
2931
+ <dim>10</dim>
2932
+ <dim>512</dim>
2933
+ </port>
2934
+ </output>
2935
+ </layer>
2936
+ <layer id="152" name="fc.weight" type="Convert" version="opset1">
2937
+ <data destination_type="f32" />
2938
+ <rt_info>
2939
+ <attribute name="decompression" version="0" />
2940
+ </rt_info>
2941
+ <input>
2942
+ <port id="0" precision="FP16">
2943
+ <dim>10</dim>
2944
+ <dim>512</dim>
2945
+ </port>
2946
+ </input>
2947
+ <output>
2948
+ <port id="1" precision="FP32">
2949
+ <dim>10</dim>
2950
+ <dim>512</dim>
2951
+ </port>
2952
+ </output>
2953
+ </layer>
2954
+ <layer id="153" name="/fc/Gemm/WithoutBiases" type="MatMul" version="opset1">
2955
+ <data transpose_a="false" transpose_b="true" />
2956
+ <input>
2957
+ <port id="0" precision="FP32">
2958
+ <dim>-1</dim>
2959
+ <dim>512</dim>
2960
+ </port>
2961
+ <port id="1" precision="FP32">
2962
+ <dim>10</dim>
2963
+ <dim>512</dim>
2964
+ </port>
2965
+ </input>
2966
+ <output>
2967
+ <port id="2" precision="FP32">
2968
+ <dim>-1</dim>
2969
+ <dim>10</dim>
2970
+ </port>
2971
+ </output>
2972
+ </layer>
2973
+ <layer id="154" name="Constant_2852" type="Const" version="opset1">
2974
+ <data element_type="f32" shape="1, 10" offset="22353696" size="40" />
2975
+ <output>
2976
+ <port id="0" precision="FP32">
2977
+ <dim>1</dim>
2978
+ <dim>10</dim>
2979
+ </port>
2980
+ </output>
2981
+ </layer>
2982
+ <layer id="155" name="logits" type="Add" version="opset1">
2983
+ <data auto_broadcast="numpy" />
2984
+ <input>
2985
+ <port id="0" precision="FP32">
2986
+ <dim>-1</dim>
2987
+ <dim>10</dim>
2988
+ </port>
2989
+ <port id="1" precision="FP32">
2990
+ <dim>1</dim>
2991
+ <dim>10</dim>
2992
+ </port>
2993
+ </input>
2994
+ <output>
2995
+ <port id="2" precision="FP32" names="logits">
2996
+ <dim>-1</dim>
2997
+ <dim>10</dim>
2998
+ </port>
2999
+ </output>
3000
+ </layer>
3001
+ <layer id="156" name="logits/sink_port_0" type="Result" version="opset1" output_names="logits">
3002
+ <input>
3003
+ <port id="0" precision="FP32">
3004
+ <dim>-1</dim>
3005
+ <dim>10</dim>
3006
+ </port>
3007
+ </input>
3008
+ </layer>
3009
+ </layers>
3010
+ <edges>
3011
+ <edge from-layer="0" from-port="0" to-layer="3" to-port="0" />
3012
+ <edge from-layer="1" from-port="0" to-layer="2" to-port="0" />
3013
+ <edge from-layer="2" from-port="1" to-layer="3" to-port="1" />
3014
+ <edge from-layer="3" from-port="2" to-layer="6" to-port="0" />
3015
+ <edge from-layer="4" from-port="0" to-layer="5" to-port="0" />
3016
+ <edge from-layer="5" from-port="1" to-layer="6" to-port="1" />
3017
+ <edge from-layer="6" from-port="2" to-layer="7" to-port="0" />
3018
+ <edge from-layer="7" from-port="1" to-layer="8" to-port="0" />
3019
+ <edge from-layer="8" from-port="1" to-layer="11" to-port="0" />
3020
+ <edge from-layer="8" from-port="1" to-layer="22" to-port="1" />
3021
+ <edge from-layer="9" from-port="0" to-layer="10" to-port="0" />
3022
+ <edge from-layer="10" from-port="1" to-layer="11" to-port="1" />
3023
+ <edge from-layer="11" from-port="2" to-layer="14" to-port="0" />
3024
+ <edge from-layer="12" from-port="0" to-layer="13" to-port="0" />
3025
+ <edge from-layer="13" from-port="1" to-layer="14" to-port="1" />
3026
+ <edge from-layer="14" from-port="2" to-layer="15" to-port="0" />
3027
+ <edge from-layer="15" from-port="1" to-layer="18" to-port="0" />
3028
+ <edge from-layer="16" from-port="0" to-layer="17" to-port="0" />
3029
+ <edge from-layer="17" from-port="1" to-layer="18" to-port="1" />
3030
+ <edge from-layer="18" from-port="2" to-layer="21" to-port="0" />
3031
+ <edge from-layer="19" from-port="0" to-layer="20" to-port="0" />
3032
+ <edge from-layer="20" from-port="1" to-layer="21" to-port="1" />
3033
+ <edge from-layer="21" from-port="2" to-layer="22" to-port="0" />
3034
+ <edge from-layer="22" from-port="2" to-layer="23" to-port="0" />
3035
+ <edge from-layer="23" from-port="1" to-layer="26" to-port="0" />
3036
+ <edge from-layer="23" from-port="1" to-layer="37" to-port="1" />
3037
+ <edge from-layer="24" from-port="0" to-layer="25" to-port="0" />
3038
+ <edge from-layer="25" from-port="1" to-layer="26" to-port="1" />
3039
+ <edge from-layer="26" from-port="2" to-layer="29" to-port="0" />
3040
+ <edge from-layer="27" from-port="0" to-layer="28" to-port="0" />
3041
+ <edge from-layer="28" from-port="1" to-layer="29" to-port="1" />
3042
+ <edge from-layer="29" from-port="2" to-layer="30" to-port="0" />
3043
+ <edge from-layer="30" from-port="1" to-layer="33" to-port="0" />
3044
+ <edge from-layer="31" from-port="0" to-layer="32" to-port="0" />
3045
+ <edge from-layer="32" from-port="1" to-layer="33" to-port="1" />
3046
+ <edge from-layer="33" from-port="2" to-layer="36" to-port="0" />
3047
+ <edge from-layer="34" from-port="0" to-layer="35" to-port="0" />
3048
+ <edge from-layer="35" from-port="1" to-layer="36" to-port="1" />
3049
+ <edge from-layer="36" from-port="2" to-layer="37" to-port="0" />
3050
+ <edge from-layer="37" from-port="2" to-layer="38" to-port="0" />
3051
+ <edge from-layer="38" from-port="1" to-layer="41" to-port="0" />
3052
+ <edge from-layer="38" from-port="1" to-layer="54" to-port="0" />
3053
+ <edge from-layer="39" from-port="0" to-layer="40" to-port="0" />
3054
+ <edge from-layer="40" from-port="1" to-layer="41" to-port="1" />
3055
+ <edge from-layer="41" from-port="2" to-layer="44" to-port="0" />
3056
+ <edge from-layer="42" from-port="0" to-layer="43" to-port="0" />
3057
+ <edge from-layer="43" from-port="1" to-layer="44" to-port="1" />
3058
+ <edge from-layer="44" from-port="2" to-layer="45" to-port="0" />
3059
+ <edge from-layer="45" from-port="1" to-layer="48" to-port="0" />
3060
+ <edge from-layer="46" from-port="0" to-layer="47" to-port="0" />
3061
+ <edge from-layer="47" from-port="1" to-layer="48" to-port="1" />
3062
+ <edge from-layer="48" from-port="2" to-layer="51" to-port="0" />
3063
+ <edge from-layer="49" from-port="0" to-layer="50" to-port="0" />
3064
+ <edge from-layer="50" from-port="1" to-layer="51" to-port="1" />
3065
+ <edge from-layer="51" from-port="2" to-layer="58" to-port="0" />
3066
+ <edge from-layer="52" from-port="0" to-layer="53" to-port="0" />
3067
+ <edge from-layer="53" from-port="1" to-layer="54" to-port="1" />
3068
+ <edge from-layer="54" from-port="2" to-layer="57" to-port="0" />
3069
+ <edge from-layer="55" from-port="0" to-layer="56" to-port="0" />
3070
+ <edge from-layer="56" from-port="1" to-layer="57" to-port="1" />
3071
+ <edge from-layer="57" from-port="2" to-layer="58" to-port="1" />
3072
+ <edge from-layer="58" from-port="2" to-layer="59" to-port="0" />
3073
+ <edge from-layer="59" from-port="1" to-layer="62" to-port="0" />
3074
+ <edge from-layer="59" from-port="1" to-layer="73" to-port="1" />
3075
+ <edge from-layer="60" from-port="0" to-layer="61" to-port="0" />
3076
+ <edge from-layer="61" from-port="1" to-layer="62" to-port="1" />
3077
+ <edge from-layer="62" from-port="2" to-layer="65" to-port="0" />
3078
+ <edge from-layer="63" from-port="0" to-layer="64" to-port="0" />
3079
+ <edge from-layer="64" from-port="1" to-layer="65" to-port="1" />
3080
+ <edge from-layer="65" from-port="2" to-layer="66" to-port="0" />
3081
+ <edge from-layer="66" from-port="1" to-layer="69" to-port="0" />
3082
+ <edge from-layer="67" from-port="0" to-layer="68" to-port="0" />
3083
+ <edge from-layer="68" from-port="1" to-layer="69" to-port="1" />
3084
+ <edge from-layer="69" from-port="2" to-layer="72" to-port="0" />
3085
+ <edge from-layer="70" from-port="0" to-layer="71" to-port="0" />
3086
+ <edge from-layer="71" from-port="1" to-layer="72" to-port="1" />
3087
+ <edge from-layer="72" from-port="2" to-layer="73" to-port="0" />
3088
+ <edge from-layer="73" from-port="2" to-layer="74" to-port="0" />
3089
+ <edge from-layer="74" from-port="1" to-layer="77" to-port="0" />
3090
+ <edge from-layer="74" from-port="1" to-layer="90" to-port="0" />
3091
+ <edge from-layer="75" from-port="0" to-layer="76" to-port="0" />
3092
+ <edge from-layer="76" from-port="1" to-layer="77" to-port="1" />
3093
+ <edge from-layer="77" from-port="2" to-layer="80" to-port="0" />
3094
+ <edge from-layer="78" from-port="0" to-layer="79" to-port="0" />
3095
+ <edge from-layer="79" from-port="1" to-layer="80" to-port="1" />
3096
+ <edge from-layer="80" from-port="2" to-layer="81" to-port="0" />
3097
+ <edge from-layer="81" from-port="1" to-layer="84" to-port="0" />
3098
+ <edge from-layer="82" from-port="0" to-layer="83" to-port="0" />
3099
+ <edge from-layer="83" from-port="1" to-layer="84" to-port="1" />
3100
+ <edge from-layer="84" from-port="2" to-layer="87" to-port="0" />
3101
+ <edge from-layer="85" from-port="0" to-layer="86" to-port="0" />
3102
+ <edge from-layer="86" from-port="1" to-layer="87" to-port="1" />
3103
+ <edge from-layer="87" from-port="2" to-layer="94" to-port="0" />
3104
+ <edge from-layer="88" from-port="0" to-layer="89" to-port="0" />
3105
+ <edge from-layer="89" from-port="1" to-layer="90" to-port="1" />
3106
+ <edge from-layer="90" from-port="2" to-layer="93" to-port="0" />
3107
+ <edge from-layer="91" from-port="0" to-layer="92" to-port="0" />
3108
+ <edge from-layer="92" from-port="1" to-layer="93" to-port="1" />
3109
+ <edge from-layer="93" from-port="2" to-layer="94" to-port="1" />
3110
+ <edge from-layer="94" from-port="2" to-layer="95" to-port="0" />
3111
+ <edge from-layer="95" from-port="1" to-layer="98" to-port="0" />
3112
+ <edge from-layer="95" from-port="1" to-layer="109" to-port="1" />
3113
+ <edge from-layer="96" from-port="0" to-layer="97" to-port="0" />
3114
+ <edge from-layer="97" from-port="1" to-layer="98" to-port="1" />
3115
+ <edge from-layer="98" from-port="2" to-layer="101" to-port="0" />
3116
+ <edge from-layer="99" from-port="0" to-layer="100" to-port="0" />
3117
+ <edge from-layer="100" from-port="1" to-layer="101" to-port="1" />
3118
+ <edge from-layer="101" from-port="2" to-layer="102" to-port="0" />
3119
+ <edge from-layer="102" from-port="1" to-layer="105" to-port="0" />
3120
+ <edge from-layer="103" from-port="0" to-layer="104" to-port="0" />
3121
+ <edge from-layer="104" from-port="1" to-layer="105" to-port="1" />
3122
+ <edge from-layer="105" from-port="2" to-layer="108" to-port="0" />
3123
+ <edge from-layer="106" from-port="0" to-layer="107" to-port="0" />
3124
+ <edge from-layer="107" from-port="1" to-layer="108" to-port="1" />
3125
+ <edge from-layer="108" from-port="2" to-layer="109" to-port="0" />
3126
+ <edge from-layer="109" from-port="2" to-layer="110" to-port="0" />
3127
+ <edge from-layer="110" from-port="1" to-layer="113" to-port="0" />
3128
+ <edge from-layer="110" from-port="1" to-layer="126" to-port="0" />
3129
+ <edge from-layer="111" from-port="0" to-layer="112" to-port="0" />
3130
+ <edge from-layer="112" from-port="1" to-layer="113" to-port="1" />
3131
+ <edge from-layer="113" from-port="2" to-layer="116" to-port="0" />
3132
+ <edge from-layer="114" from-port="0" to-layer="115" to-port="0" />
3133
+ <edge from-layer="115" from-port="1" to-layer="116" to-port="1" />
3134
+ <edge from-layer="116" from-port="2" to-layer="117" to-port="0" />
3135
+ <edge from-layer="117" from-port="1" to-layer="120" to-port="0" />
3136
+ <edge from-layer="118" from-port="0" to-layer="119" to-port="0" />
3137
+ <edge from-layer="119" from-port="1" to-layer="120" to-port="1" />
3138
+ <edge from-layer="120" from-port="2" to-layer="123" to-port="0" />
3139
+ <edge from-layer="121" from-port="0" to-layer="122" to-port="0" />
3140
+ <edge from-layer="122" from-port="1" to-layer="123" to-port="1" />
3141
+ <edge from-layer="123" from-port="2" to-layer="130" to-port="0" />
3142
+ <edge from-layer="124" from-port="0" to-layer="125" to-port="0" />
3143
+ <edge from-layer="125" from-port="1" to-layer="126" to-port="1" />
3144
+ <edge from-layer="126" from-port="2" to-layer="129" to-port="0" />
3145
+ <edge from-layer="127" from-port="0" to-layer="128" to-port="0" />
3146
+ <edge from-layer="128" from-port="1" to-layer="129" to-port="1" />
3147
+ <edge from-layer="129" from-port="2" to-layer="130" to-port="1" />
3148
+ <edge from-layer="130" from-port="2" to-layer="131" to-port="0" />
3149
+ <edge from-layer="131" from-port="1" to-layer="134" to-port="0" />
3150
+ <edge from-layer="131" from-port="1" to-layer="145" to-port="1" />
3151
+ <edge from-layer="132" from-port="0" to-layer="133" to-port="0" />
3152
+ <edge from-layer="133" from-port="1" to-layer="134" to-port="1" />
3153
+ <edge from-layer="134" from-port="2" to-layer="137" to-port="0" />
3154
+ <edge from-layer="135" from-port="0" to-layer="136" to-port="0" />
3155
+ <edge from-layer="136" from-port="1" to-layer="137" to-port="1" />
3156
+ <edge from-layer="137" from-port="2" to-layer="138" to-port="0" />
3157
+ <edge from-layer="138" from-port="1" to-layer="141" to-port="0" />
3158
+ <edge from-layer="139" from-port="0" to-layer="140" to-port="0" />
3159
+ <edge from-layer="140" from-port="1" to-layer="141" to-port="1" />
3160
+ <edge from-layer="141" from-port="2" to-layer="144" to-port="0" />
3161
+ <edge from-layer="142" from-port="0" to-layer="143" to-port="0" />
3162
+ <edge from-layer="143" from-port="1" to-layer="144" to-port="1" />
3163
+ <edge from-layer="144" from-port="2" to-layer="145" to-port="0" />
3164
+ <edge from-layer="145" from-port="2" to-layer="146" to-port="0" />
3165
+ <edge from-layer="146" from-port="1" to-layer="148" to-port="0" />
3166
+ <edge from-layer="147" from-port="0" to-layer="148" to-port="1" />
3167
+ <edge from-layer="148" from-port="2" to-layer="150" to-port="0" />
3168
+ <edge from-layer="149" from-port="0" to-layer="150" to-port="1" />
3169
+ <edge from-layer="150" from-port="2" to-layer="153" to-port="0" />
3170
+ <edge from-layer="151" from-port="0" to-layer="152" to-port="0" />
3171
+ <edge from-layer="152" from-port="1" to-layer="153" to-port="1" />
3172
+ <edge from-layer="153" from-port="2" to-layer="155" to-port="0" />
3173
+ <edge from-layer="154" from-port="0" to-layer="155" to-port="1" />
3174
+ <edge from-layer="155" from-port="2" to-layer="156" to-port="0" />
3175
+ </edges>
3176
+ <rt_info>
3177
+ <info name="OpenVINO Runtime" value="2026.2.1-21919-ede283a88e3-releases/2026/2" />
3178
+ <Runtime_version value="2026.2.1-21919-ede283a88e3-releases/2026/2" />
3179
+ <conversion_parameters>
3180
+ <input_model value="DIR/animal-cls-v1.onnx" />
3181
+ <is_python_object value="False" />
3182
+ </conversion_parameters>
3183
+ <framework>
3184
+ <labels value="[&quot;butterfly&quot;, &quot;cat&quot;, &quot;chicken&quot;, &quot;cow&quot;, &quot;dog&quot;, &quot;elephant&quot;, &quot;horse&quot;, &quot;sheep&quot;, &quot;spider&quot;, &quot;squirrel&quot;]" />
3185
+ <preprocess value="imagenet_norm;224;RGB;NCHW" />
3186
+ </framework>
3187
+ </rt_info>
3188
+ </net>