qaihm-bot commited on
Commit
ffd36b8
·
verified ·
1 Parent(s): 984e627

See https://github.com/quic/ai-hub-models/releases/v0.42.0 for changelog.

Inception-v3_float.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e54f3da123c8c9532f6f015d55f89670ebf992b614a71b8ec5b092d8ffd325f
3
- size 95526996
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9dc65ceb5f1ead7b3f9f7e009be7874074ddf3c0b20d9c446b50e8b77cdafc9c
3
+ size 95527100
Inception-v3_float.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:adeefff97bb5dc81e9d768cc3ab16370ea8001fa938f08b8ee3e2e8ceb0d6ddd
3
  size 88766454
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6466ab647e5125df104bbbfb24592d5adbf5f54cbb308f76880e91269c6c051c
3
  size 88766454
Inception-v3_w8a8.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:380494682c7f65b420f259981d2780b6893903caffbce66d8f1036773ec76315
3
- size 25105292
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2179d0350244bddd7be5b345b5212735797487eadb2fbb673c6ec58b71f5ed90
3
+ size 25105396
Inception-v3_w8a8.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8b83636f86b253f510b2e02da33cc590b8cdade939ed9d7eadf531efd0badea6
3
- size 41143725
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:658650614a6238989bab088c41c74220af5ba166e5d4018388c50ff7121eae5e
3
+ size 41143727
Inception-v3_w8a8.tflite CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:963d7c6b22ed54f38d95b4095e2909642cfe001e2677a0bf95e707d1ee0c689c
3
  size 24447896
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7d0f0d4aa9c997c7207bf065326c40c2e1ddc06fcabbb0101f4267d53dfaac0
3
  size 24447896
README.md CHANGED
@@ -37,71 +37,74 @@ More details on model performance across various devices, can be found
37
 
38
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
39
  |---|---|---|---|---|---|---|---|---|
40
- | Inception-v3 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 7.729 ms | 0 - 60 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
41
- | Inception-v3 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 7.763 ms | 0 - 24 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
42
- | Inception-v3 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2.109 ms | 0 - 98 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
43
- | Inception-v3 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 2.332 ms | 0 - 39 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
44
- | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.307 ms | 0 - 369 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
45
- | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.358 ms | 0 - 150 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
46
- | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 1.698 ms | 1 - 169 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
47
- | Inception-v3 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.181 ms | 0 - 60 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
48
- | Inception-v3 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.22 ms | 1 - 25 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
49
- | Inception-v3 | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 7.729 ms | 0 - 60 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
50
- | Inception-v3 | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 7.763 ms | 0 - 24 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
51
- | Inception-v3 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1.305 ms | 0 - 364 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
52
- | Inception-v3 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 1.359 ms | 0 - 130 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
53
- | Inception-v3 | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 2.586 ms | 0 - 65 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
54
- | Inception-v3 | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 2.656 ms | 1 - 28 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
55
- | Inception-v3 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 1.305 ms | 0 - 364 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
56
- | Inception-v3 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 1.358 ms | 0 - 146 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
57
- | Inception-v3 | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.181 ms | 0 - 60 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
58
- | Inception-v3 | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.22 ms | 1 - 25 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
59
- | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.973 ms | 0 - 96 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
60
- | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.991 ms | 0 - 33 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
61
- | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.19 ms | 0 - 35 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
62
- | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.837 ms | 0 - 66 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
63
- | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.862 ms | 1 - 31 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
64
- | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 1.025 ms | 0 - 29 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
65
- | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.765 ms | 0 - 66 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
66
- | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.758 ms | 1 - 31 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
67
- | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 0.958 ms | 0 - 30 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
68
- | Inception-v3 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.383 ms | 165 - 165 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
69
- | Inception-v3 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.539 ms | 46 - 46 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
70
- | Inception-v3 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 1.508 ms | 0 - 38 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
71
- | Inception-v3 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 1.49 ms | 0 - 41 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
72
- | Inception-v3 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 0.763 ms | 0 - 55 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
73
- | Inception-v3 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 0.89 ms | 0 - 59 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
74
- | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 0.644 ms | 0 - 135 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
75
- | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.591 ms | 0 - 144 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
76
- | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 0.88 ms | 0 - 108 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
77
- | Inception-v3 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 0.816 ms | 0 - 38 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
78
- | Inception-v3 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 0.758 ms | 0 - 41 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
79
- | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 39.915 ms | 1 - 36 MB | GPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
80
- | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 2.784 ms | 0 - 53 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
81
- | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 20.26 ms | 16 - 33 MB | CPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
82
- | Inception-v3 | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 8.45 ms | 0 - 3 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
83
- | Inception-v3 | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 18.337 ms | 8 - 37 MB | CPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
84
- | Inception-v3 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 1.508 ms | 0 - 38 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
85
- | Inception-v3 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 1.49 ms | 0 - 41 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
86
- | Inception-v3 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 0.638 ms | 0 - 6 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
87
- | Inception-v3 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.596 ms | 0 - 143 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
88
- | Inception-v3 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 1.12 ms | 0 - 44 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
89
- | Inception-v3 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.138 ms | 0 - 47 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
90
- | Inception-v3 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 0.636 ms | 0 - 133 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
91
- | Inception-v3 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.594 ms | 0 - 142 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
92
- | Inception-v3 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 0.816 ms | 0 - 38 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
93
- | Inception-v3 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 0.758 ms | 0 - 41 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
94
- | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.466 ms | 0 - 59 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
95
- | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.467 ms | 0 - 61 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
96
- | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 0.652 ms | 0 - 63 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
97
- | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.425 ms | 0 - 44 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
98
- | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.394 ms | 0 - 46 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
99
- | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 0.56 ms | 0 - 54 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
100
- | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.399 ms | 0 - 44 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
101
- | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.362 ms | 0 - 47 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
102
- | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 0.578 ms | 0 - 53 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
103
- | Inception-v3 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 0.654 ms | 132 - 132 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
104
- | Inception-v3 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.745 ms | 26 - 26 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
 
 
 
105
 
106
 
107
 
@@ -115,9 +118,9 @@ pip install qai-hub-models
115
  ```
116
 
117
 
118
- ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
119
 
120
- Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
121
  Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
122
 
123
  With this API token, you can configure your client to run models on the cloud
@@ -125,7 +128,7 @@ hosted devices.
125
  ```bash
126
  qai-hub configure --api_token API_TOKEN
127
  ```
128
- Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
129
 
130
 
131
 
@@ -236,7 +239,7 @@ With the output of the model, you can compute like PSNR, relative errors or
236
  spot check the output with expected output.
237
 
238
  **Note**: This on-device profiling and inference requires access to Qualcomm®
239
- AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
240
 
241
 
242
 
 
37
 
38
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
39
  |---|---|---|---|---|---|---|---|---|
40
+ | Inception-v3 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 7.745 ms | 0 - 61 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
41
+ | Inception-v3 | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 7.744 ms | 0 - 26 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
42
+ | Inception-v3 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2.116 ms | 0 - 101 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
43
+ | Inception-v3 | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 2.322 ms | 0 - 38 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
44
+ | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.311 ms | 0 - 365 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
45
+ | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.356 ms | 0 - 131 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
46
+ | Inception-v3 | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 1.694 ms | 0 - 143 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
47
+ | Inception-v3 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.231 ms | 0 - 61 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
48
+ | Inception-v3 | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.205 ms | 1 - 27 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
49
+ | Inception-v3 | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 7.745 ms | 0 - 61 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
50
+ | Inception-v3 | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 7.744 ms | 0 - 26 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
51
+ | Inception-v3 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1.306 ms | 0 - 367 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
52
+ | Inception-v3 | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 1.361 ms | 0 - 150 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
53
+ | Inception-v3 | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 2.58 ms | 0 - 66 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
54
+ | Inception-v3 | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 2.625 ms | 1 - 30 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
55
+ | Inception-v3 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 1.296 ms | 0 - 363 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
56
+ | Inception-v3 | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 1.359 ms | 0 - 151 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
57
+ | Inception-v3 | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.231 ms | 0 - 61 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
58
+ | Inception-v3 | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.205 ms | 1 - 27 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
59
+ | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.969 ms | 0 - 100 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
60
+ | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.998 ms | 0 - 34 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
61
+ | Inception-v3 | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.186 ms | 0 - 38 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
62
+ | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.837 ms | 0 - 68 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
63
+ | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.864 ms | 1 - 33 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
64
+ | Inception-v3 | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 1.026 ms | 0 - 30 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
65
+ | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.765 ms | 0 - 67 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.tflite) |
66
+ | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.762 ms | 1 - 31 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
67
+ | Inception-v3 | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 0.936 ms | 0 - 30 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
68
+ | Inception-v3 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.412 ms | 145 - 145 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.dlc) |
69
+ | Inception-v3 | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.522 ms | 46 - 46 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3.onnx.zip) |
70
+ | Inception-v3 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 1.498 ms | 0 - 39 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
71
+ | Inception-v3 | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 1.523 ms | 0 - 42 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
72
+ | Inception-v3 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 0.806 ms | 0 - 55 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
73
+ | Inception-v3 | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 0.903 ms | 0 - 56 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
74
+ | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 0.639 ms | 0 - 136 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
75
+ | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.598 ms | 0 - 144 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
76
+ | Inception-v3 | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 0.882 ms | 0 - 129 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
77
+ | Inception-v3 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.834 ms | 0 - 39 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
78
+ | Inception-v3 | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.627 ms | 0 - 43 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
79
+ | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 2.534 ms | 0 - 58 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
80
+ | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 2.859 ms | 0 - 55 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
81
+ | Inception-v3 | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 20.29 ms | 16 - 33 MB | CPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
82
+ | Inception-v3 | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 8.3 ms | 0 - 2 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
83
+ | Inception-v3 | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 18.074 ms | 12 - 40 MB | CPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
84
+ | Inception-v3 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 1.498 ms | 0 - 39 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
85
+ | Inception-v3 | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 1.523 ms | 0 - 42 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
86
+ | Inception-v3 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 0.643 ms | 0 - 136 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
87
+ | Inception-v3 | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.599 ms | 0 - 141 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
88
+ | Inception-v3 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 1.118 ms | 0 - 45 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
89
+ | Inception-v3 | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.122 ms | 0 - 48 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
90
+ | Inception-v3 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 0.643 ms | 0 - 7 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
91
+ | Inception-v3 | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.599 ms | 0 - 144 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
92
+ | Inception-v3 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.834 ms | 0 - 39 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
93
+ | Inception-v3 | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.627 ms | 0 - 43 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
94
+ | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.478 ms | 0 - 54 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
95
+ | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.468 ms | 0 - 52 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
96
+ | Inception-v3 | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 0.667 ms | 0 - 62 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
97
+ | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.425 ms | 0 - 50 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
98
+ | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.397 ms | 0 - 54 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
99
+ | Inception-v3 | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 0.559 ms | 0 - 52 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
100
+ | Inception-v3 | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | TFLITE | 0.975 ms | 0 - 50 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
101
+ | Inception-v3 | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | QNN_DLC | 0.947 ms | 0 - 52 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
102
+ | Inception-v3 | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 18.738 ms | 18 - 35 MB | CPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
103
+ | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.4 ms | 0 - 44 MB | NPU | [Inception-v3.tflite](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.tflite) |
104
+ | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.369 ms | 0 - 45 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
105
+ | Inception-v3 | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 0.568 ms | 0 - 51 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
106
+ | Inception-v3 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 0.656 ms | 133 - 133 MB | NPU | [Inception-v3.dlc](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.dlc) |
107
+ | Inception-v3 | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.749 ms | 26 - 26 MB | NPU | [Inception-v3.onnx.zip](https://huggingface.co/qualcomm/Inception-v3/blob/main/Inception-v3_w8a8.onnx.zip) |
108
 
109
 
110
 
 
118
  ```
119
 
120
 
121
+ ## Configure Qualcomm® AI Hub Workbench to run this model on a cloud-hosted device
122
 
123
+ Sign-in to [Qualcomm® AI Hub Workbench](https://workbench.aihub.qualcomm.com/) with your
124
  Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
125
 
126
  With this API token, you can configure your client to run models on the cloud
 
128
  ```bash
129
  qai-hub configure --api_token API_TOKEN
130
  ```
131
+ Navigate to [docs](https://workbench.aihub.qualcomm.com/docs/) for more information.
132
 
133
 
134
 
 
239
  spot check the output with expected output.
240
 
241
  **Note**: This on-device profiling and inference requires access to Qualcomm®
242
+ AI Hub Workbench. [Sign up for access](https://myaccount.qualcomm.com/signup).
243
 
244
 
245