Hugging Face's logo

qualcomm
/
VIT

Image Classification
PyTorch
LiteRT
backbone
android
Model card Files
xet
Community
qaihm-bot commited on
Commit
16ec8ae
·
verified ·
1 Parent(s): e65365d

v0.38.0

Browse files

See https://github.com/quic/ai-hub-models/releases/v0.38.0 for changelog.

Files changed (17) hide show
  1. README.md +81 -90
  2. VIT_float.dlc +2 -2
  3. VIT_float.onnx.zip +2 -2
  4. VIT_w8a16.dlc +2 -2
  5. VIT_w8a16.onnx.zip +2 -2
  6. VIT_w8a8.onnx.zip +2 -2
  7. VIT_w8a8.tflite +1 -1
  8. VIT_w8a8_mixed_int16.dlc +2 -2
  9. VIT_w8a8_mixed_int16.onnx.zip +2 -2
  10. precompiled/qualcomm-qcs6490-proxy/VIT_w8a16.bin +0 -3
  11. precompiled/qualcomm-qcs6490-proxy/tool-versions.yaml +0 -3
  12. precompiled/qualcomm-snapdragon-x-elite/VIT_float.bin +0 -3
  13. precompiled/qualcomm-snapdragon-x-elite/VIT_float.onnx.zip +0 -3
  14. precompiled/qualcomm-snapdragon-x-elite/VIT_w8a16.bin +0 -3
  15. precompiled/qualcomm-snapdragon-x-elite/VIT_w8a8_mixed_int16.bin +0 -3
  16. precompiled/qualcomm-snapdragon-x-elite/tool-versions.yaml +0 -3
  17. tool-versions.yaml +1 -1
README.md CHANGED
@@ -38,95 +38,86 @@ More details on model performance across various devices, can be found
38
 
39
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
- | VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 42.876 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
42
- | VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 45.163 ms | 0 - 320 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
43
- | VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 17.073 ms | 0 - 299 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
44
- | VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 21.48 ms | 1 - 324 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
45
- | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 12.48 ms | 0 - 23 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
46
- | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 13.839 ms | 0 - 31 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
47
- | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 13.477 ms | 0 - 23 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
48
- | VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 15.25 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
49
- | VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 16.627 ms | 1 - 327 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
50
- | VIT | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 42.876 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
51
- | VIT | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 45.163 ms | 0 - 320 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
52
- | VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 12.452 ms | 0 - 16 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
53
- | VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 13.834 ms | 0 - 33 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
54
- | VIT | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 19.267 ms | 0 - 290 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
55
- | VIT | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 19.77 ms | 0 - 320 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
56
- | VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 12.492 ms | 0 - 14 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
57
- | VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 13.835 ms | 0 - 31 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
58
- | VIT | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 15.25 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
59
- | VIT | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 16.627 ms | 1 - 327 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
60
- | VIT | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 12.462 ms | 0 - 20 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
61
- | VIT | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 13.849 ms | 0 - 29 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
62
- | VIT | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 13.325 ms | 1 - 25 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
63
- | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 8.515 ms | 0 - 311 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
64
- | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 9.54 ms | 1 - 324 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
65
- | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.146 ms | 0 - 323 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
66
- | VIT | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 7.282 ms | 0 - 309 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
67
- | VIT | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 7.977 ms | 1 - 313 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
68
- | VIT | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 7.606 ms | 1 - 323 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
69
- | VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 14.618 ms | 1094 - 1094 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
70
- | VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 13.843 ms | 171 - 171 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
71
- | VIT | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 65.524 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
72
- | VIT | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 50.065 ms | 0 - 223 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
73
- | VIT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 26.186 ms | 0 - 47 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
74
- | VIT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 168.292 ms | 536 - 787 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
75
- | VIT | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 23.047 ms | 0 - 197 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
76
- | VIT | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 196.718 ms | 0 - 1520 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
77
- | VIT | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 536.174 ms | 70 - 87 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
78
- | VIT | w8a16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 631.549 ms | 44 - 129 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
79
- | VIT | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 65.524 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
80
- | VIT | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 26.173 ms | 0 - 48 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
81
- | VIT | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 36.991 ms | 0 - 215 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
82
- | VIT | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 26.05 ms | 0 - 48 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
83
- | VIT | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 23.047 ms | 0 - 197 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
84
- | VIT | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 26.057 ms | 0 - 48 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
85
- | VIT | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 159.216 ms | 653 - 900 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
86
- | VIT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 19.859 ms | 0 - 201 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
87
- | VIT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 130.896 ms | 663 - 829 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
88
- | VIT | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 16.724 ms | 0 - 188 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
89
- | VIT | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 106.079 ms | 692 - 841 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
90
- | VIT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 25.769 ms | 325 - 325 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
91
- | VIT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 156.58 ms | 926 - 926 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
92
- | VIT | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 15.863 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
93
- | VIT | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 8.21 ms | 0 - 56 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
94
- | VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 7.618 ms | 0 - 22 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
95
- | VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 159.971 ms | 662 - 842 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
96
- | VIT | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 8.002 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
97
- | VIT | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 62.138 ms | 2 - 45 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
98
- | VIT | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 446.395 ms | 29 - 47 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
99
- | VIT | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 467.202 ms | 29 - 83 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
100
- | VIT | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 15.863 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
101
- | VIT | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 7.628 ms | 0 - 21 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
102
- | VIT | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 9.885 ms | 0 - 49 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
103
- | VIT | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 7.618 ms | 0 - 81 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
104
- | VIT | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 8.002 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
105
- | VIT | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 7.663 ms | 0 - 80 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
106
- | VIT | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 151.575 ms | 598 - 837 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
107
- | VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.385 ms | 0 - 53 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
108
- | VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 127.612 ms | 675 - 832 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
109
- | VIT | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 4.991 ms | 0 - 56 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
110
- | VIT | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 96.727 ms | 668 - 803 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
111
- | VIT | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 144.921 ms | 926 - 926 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
112
- | VIT | w8a8_mixed_int16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 53.742 ms | 0 - 242 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
113
- | VIT | w8a8_mixed_int16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 21.16 ms | 3 - 44 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
114
- | VIT | w8a8_mixed_int16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 185.126 ms | 479 - 727 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
115
- | VIT | w8a8_mixed_int16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 20.135 ms | 0 - 229 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
116
- | VIT | w8a8_mixed_int16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 459.794 ms | 47 - 65 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
117
- | VIT | w8a8_mixed_int16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 527.283 ms | 48 - 62 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
118
- | VIT | w8a8_mixed_int16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 53.742 ms | 0 - 242 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
119
- | VIT | w8a8_mixed_int16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 21.252 ms | 0 - 41 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
120
- | VIT | w8a8_mixed_int16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 21.119 ms | 0 - 41 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
121
- | VIT | w8a8_mixed_int16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 20.135 ms | 0 - 229 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
122
- | VIT | w8a8_mixed_int16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 21.121 ms | 0 - 41 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
123
- | VIT | w8a8_mixed_int16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 194.107 ms | 524 - 807 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
124
- | VIT | w8a8_mixed_int16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 16.395 ms | 3 - 252 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
125
- | VIT | w8a8_mixed_int16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 172.624 ms | 541 - 721 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
126
- | VIT | w8a8_mixed_int16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 12.962 ms | 0 - 260 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
127
- | VIT | w8a8_mixed_int16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 119.534 ms | 557 - 737 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
128
- | VIT | w8a8_mixed_int16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 20.932 ms | 398 - 398 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
129
- | VIT | w8a8_mixed_int16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 170.989 ms | 922 - 922 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
130
 
131
 
132
 
@@ -208,7 +199,7 @@ from qai_hub_models.models.vit import Model
208
  torch_model = Model.from_pretrained()
209
 
210
  # Device
211
- device = hub.Device("Samsung Galaxy S24")
212
 
213
  # Trace model
214
  input_shape = torch_model.get_input_spec()
 
38
 
39
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
40
  |---|---|---|---|---|---|---|---|---|
41
+ | VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 42.829 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
42
+ | VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 44.882 ms | 1 - 328 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
43
+ | VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 17.158 ms | 0 - 301 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
44
+ | VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 21.72 ms | 1 - 318 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
45
+ | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 12.547 ms | 0 - 16 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
46
+ | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 14.02 ms | 0 - 31 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
47
+ | VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 13.399 ms | 0 - 219 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
48
+ | VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 15.223 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
49
+ | VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 16.519 ms | 1 - 331 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
50
+ | VIT | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 42.829 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
51
+ | VIT | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 44.882 ms | 1 - 328 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
52
+ | VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 12.544 ms | 0 - 14 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
53
+ | VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 14.059 ms | 0 - 28 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
54
+ | VIT | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 19.329 ms | 0 - 290 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
55
+ | VIT | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 19.721 ms | 1 - 324 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
56
+ | VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 12.539 ms | 0 - 15 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
57
+ | VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 14.097 ms | 0 - 31 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
58
+ | VIT | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 15.223 ms | 0 - 306 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
59
+ | VIT | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 16.519 ms | 1 - 331 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
60
+ | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 8.453 ms | 0 - 312 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
61
+ | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 9.524 ms | 0 - 336 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
62
+ | VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.118 ms | 0 - 331 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
63
+ | VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 6.18 ms | 0 - 310 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT.tflite) |
64
+ | VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 6.952 ms | 1 - 324 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
65
+ | VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 6.329 ms | 0 - 325 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
66
+ | VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 14.6 ms | 1085 - 1085 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT.dlc) |
67
+ | VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 13.833 ms | 171 - 171 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT.onnx.zip) |
68
+ | VIT | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 65.394 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
69
+ | VIT | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 54.53 ms | 0 - 226 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
70
+ | VIT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 25.844 ms | 0 - 49 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
71
+ | VIT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 158.841 ms | 510 - 727 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
72
+ | VIT | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 22.973 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
73
+ | VIT | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 196.519 ms | 0 - 1574 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
74
+ | VIT | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 589.625 ms | 71 - 88 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
75
+ | VIT | w8a16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 570.422 ms | 47 - 129 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
76
+ | VIT | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 65.394 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
77
+ | VIT | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 25.808 ms | 0 - 48 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
78
+ | VIT | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 36.921 ms | 0 - 215 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
79
+ | VIT | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 25.705 ms | 0 - 48 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
80
+ | VIT | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 22.973 ms | 0 - 196 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
81
+ | VIT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 19.675 ms | 0 - 207 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
82
+ | VIT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 131.918 ms | 672 - 857 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
83
+ | VIT | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 15.34 ms | 0 - 195 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
84
+ | VIT | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 104.61 ms | 681 - 848 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
85
+ | VIT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 26.079 ms | 313 - 313 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.dlc) |
86
+ | VIT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 156.034 ms | 924 - 924 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a16.onnx.zip) |
87
+ | VIT | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 15.947 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
88
+ | VIT | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 8.357 ms | 0 - 57 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
89
+ | VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 7.602 ms | 0 - 31 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
90
+ | VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 160.338 ms | 512 - 753 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
91
+ | VIT | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 7.985 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
92
+ | VIT | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 97.987 ms | 2 - 45 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
93
+ | VIT | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 439.547 ms | 28 - 46 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
94
+ | VIT | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 452.699 ms | 28 - 93 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
95
+ | VIT | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 15.947 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
96
+ | VIT | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 7.622 ms | 0 - 22 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
97
+ | VIT | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 9.962 ms | 0 - 50 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
98
+ | VIT | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 7.651 ms | 0 - 22 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
99
+ | VIT | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 7.985 ms | 0 - 47 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
100
+ | VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.369 ms | 0 - 55 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
101
+ | VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 132.271 ms | 674 - 852 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
102
+ | VIT | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 4.211 ms | 0 - 56 MB | NPU | [VIT.tflite](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.tflite) |
103
+ | VIT | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 112.83 ms | 671 - 817 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
104
+ | VIT | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 164.811 ms | 926 - 926 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8.onnx.zip) |
105
+ | VIT | w8a8_mixed_int16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 54.046 ms | 0 - 250 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
106
+ | VIT | w8a8_mixed_int16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 21.222 ms | 0 - 41 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
107
+ | VIT | w8a8_mixed_int16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 197.634 ms | 525 - 775 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
108
+ | VIT | w8a8_mixed_int16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 20.099 ms | 0 - 236 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
109
+ | VIT | w8a8_mixed_int16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 457.613 ms | 61 - 79 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
110
+ | VIT | w8a8_mixed_int16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 452.815 ms | 49 - 109 MB | CPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
111
+ | VIT | w8a8_mixed_int16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 54.046 ms | 0 - 250 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
112
+ | VIT | w8a8_mixed_int16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 21.228 ms | 0 - 41 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
113
+ | VIT | w8a8_mixed_int16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 21.122 ms | 0 - 42 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
114
+ | VIT | w8a8_mixed_int16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 20.099 ms | 0 - 236 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
115
+ | VIT | w8a8_mixed_int16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 16.26 ms | 0 - 245 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
116
+ | VIT | w8a8_mixed_int16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 158.987 ms | 542 - 785 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
117
+ | VIT | w8a8_mixed_int16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 13.207 ms | 0 - 256 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
118
+ | VIT | w8a8_mixed_int16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 118.795 ms | 551 - 730 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
119
+ | VIT | w8a8_mixed_int16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 20.947 ms | 348 - 348 MB | NPU | [VIT.dlc](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.dlc) |
120
+ | VIT | w8a8_mixed_int16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 198.967 ms | 926 - 926 MB | NPU | [VIT.onnx.zip](https://huggingface.co/qualcomm/VIT/blob/main/VIT_w8a8_mixed_int16.onnx.zip) |
 
 
 
 
 
 
 
 
 
121
 
122
 
123
 
 
199
  torch_model = Model.from_pretrained()
200
 
201
  # Device
202
+ device = hub.Device("Samsung Galaxy S25")
203
 
204
  # Trace model
205
  input_shape = torch_model.get_input_spec()
VIT_float.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:32ddf641123ff043955ed19475a094eb64a20bd477ebc3abe34024ed9d8b25d6
3
- size 347071820
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6eb6ff4c71de29eb824af7d643a0678f411c87d21309fd0fb2cd03888bb62a6a
3
+ size 347071900
VIT_float.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d35b235ba319438f309c68e1b2703506fe31c3fc40c626541d47f20b6f82045b
3
- size 322933270
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fec5c5c6c186191c3bd476b799fe2bb7130c49ef3e9d15d6cc4d31599af31ffe
3
+ size 322933371
VIT_w8a16.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:01039ccad9157496c610986289351b529a41cf8ce8b1d4e064605a716e677a1a
3
- size 90408500
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0de815c3034dda2b7ee2ca586f9bd7c1a92b59bbf4f05a1cdbcb13760937402
3
+ size 90408516
VIT_w8a16.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:221dd37d2bca097cbf1241f5a8c1a60b461b62424565b0c090c8190aff510792
3
- size 304716748
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:948734e68c7013bb3b1bd48b64c98cd8e4a0dc08e6b17eb5cab118703cc3d1bd
3
+ size 304716855
VIT_w8a8.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3932d403002e3dcf64268638eb30049a5e4096e13a771dec866254ca1436c519
3
- size 304768370
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0252282fd80291adede005245bc3023b1d7abed06bc4515af14d5685c747eafb
3
+ size 304803023
VIT_w8a8.tflite CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:eef3ec1f22ed0c794ec773252ad7f1bf2226e58ba0370f18c0f7a0343c5b5d5f
3
  size 87210560
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bff4796af741d675cb381b2f37517844b89fb65cda0448dc38ce8d3ef34d759
3
  size 87210560
VIT_w8a8_mixed_int16.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0759afa64986be7cd7f11c8bbea603e1c00824a1c84669ed999b3af0ff962ff7
3
- size 95143212
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:296358ed8e9bcb3bd90d0a471d30dae8c4aa709a4cb7a79e6548ec567aa13dc9
3
+ size 95147916
VIT_w8a8_mixed_int16.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e84174c3d970af02fb179343223823f81483da59f43da33f45d7ffbfef532e1
3
- size 307374249
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:257564a696cc5b1f17ba0d5a38df86fee3e4b065afc8c0915087daa6d32f1e18
3
+ size 307374355
precompiled/qualcomm-qcs6490-proxy/VIT_w8a16.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:036186b7941f38dbb9bc89dd4d6233019f07c8f9c5d92b22feaf96bd830b21a5
3
- size 95862784
 
 
 
 
precompiled/qualcomm-qcs6490-proxy/tool-versions.yaml DELETED
@@ -1,3 +0,0 @@
1
- tool_versions:
2
- qnn_context_binary:
3
- qairt: 2.37.0.250724175447_124859
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/VIT_float.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:c9e5cdb0e4d852538a00896ca0206206170045c599ca07cea6deb22e2ed5d4ed
3
- size 178946048
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/VIT_float.onnx.zip DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:1e2767c09e6c58a2ed6b496e44ba61def5207bb475f5812e9842e7fbab538216
3
- size 162519787
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/VIT_w8a16.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:469f1a75c5632dec00d997643f3a5f4f77eb464169bdeb3bf640fea14d1861a2
3
- size 90537984
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/VIT_w8a8_mixed_int16.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6236e26e78ae02e1c632963067f413e8bba686ed3b5571bf35ab0a2ed73b1260
3
- size 94748672
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/tool-versions.yaml DELETED
@@ -1,3 +0,0 @@
1
- tool_versions:
2
- qnn_context_binary:
3
- qairt: 2.37.0.250724175447_124859
 
 
 
 
tool-versions.yaml CHANGED
@@ -1,4 +1,4 @@
1
  tool_versions:
2
  onnx:
3
- qairt: 2.36.4.250725200057_123280
4
  onnx_runtime: 1.22.2
 
1
  tool_versions:
2
  onnx:
3
+ qairt: 2.37.1.250807093845_124904
4
  onnx_runtime: 1.22.2