qaihm-bot commited on
Commit
3e85e30
·
verified ·
1 Parent(s): e05b295

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

README.md CHANGED
@@ -37,57 +37,52 @@ 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
- | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 44.225 ms | 0 - 267 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
41
- | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 38.383 ms | 1 - 511 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
42
- | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 23.323 ms | 0 - 259 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
43
- | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 24.187 ms | 1 - 235 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
44
- | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 18.459 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
45
- | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 15.738 ms | 0 - 60 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
46
- | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 15.905 ms | 3 - 52 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
47
- | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 20.523 ms | 0 - 268 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
48
- | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 17.809 ms | 1 - 533 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
49
- | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 44.225 ms | 0 - 267 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
50
- | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 38.383 ms | 1 - 511 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
51
- | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 18.539 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
52
- | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 15.784 ms | 0 - 57 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
53
- | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 26.394 ms | 0 - 259 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
54
- | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 23.312 ms | 1 - 510 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
55
- | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 18.499 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
56
- | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 15.739 ms | 0 - 57 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
57
- | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 20.523 ms | 0 - 268 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
58
- | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 17.809 ms | 1 - 533 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
59
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 18.581 ms | 0 - 30 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
60
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 15.742 ms | 0 - 57 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
61
- | Swin-Small | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 15.8 ms | 1 - 54 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
62
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 12.422 ms | 0 - 267 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
63
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 10.564 ms | 1 - 741 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
64
- | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 10.575 ms | 1 - 731 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
65
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 12.17 ms | 0 - 259 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
66
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 7.984 ms | 1 - 529 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
67
- | Swin-Small | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 8.13 ms | 1 - 237 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
68
- | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 16.63 ms | 522 - 522 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
69
- | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 16.204 ms | 100 - 100 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
70
- | Swin-Small | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 28.633 ms | 0 - 278 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
71
- | Swin-Small | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 19.225 ms | 0 - 286 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
72
- | Swin-Small | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 15.659 ms | 0 - 77 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
73
- | Swin-Small | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 123.961 ms | 259 - 427 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
74
- | Swin-Small | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 16.087 ms | 0 - 273 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
75
- | Swin-Small | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 47.069 ms | 0 - 711 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
76
- | Swin-Small | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 343.207 ms | 66 - 92 MB | CPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
77
- | Swin-Small | w8a16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 395.829 ms | 52 - 106 MB | CPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
78
- | Swin-Small | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 28.633 ms | 0 - 278 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
79
- | Swin-Small | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 15.722 ms | 0 - 71 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
80
- | Swin-Small | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 18.51 ms | 0 - 192 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
81
- | Swin-Small | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 15.658 ms | 0 - 60 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
82
- | Swin-Small | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 16.087 ms | 0 - 273 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
83
- | Swin-Small | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 15.615 ms | 0 - 71 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
84
- | Swin-Small | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 124.312 ms | 270 - 432 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
85
- | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 10.567 ms | 0 - 294 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
86
- | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 102.449 ms | 287 - 529 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
87
- | Swin-Small | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 9.596 ms | 0 - 273 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
88
- | Swin-Small | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 64.878 ms | 287 - 526 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
89
- | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 16.508 ms | 165 - 165 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
90
- | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 99.265 ms | 460 - 460 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
91
 
92
 
93
 
@@ -169,7 +164,7 @@ from qai_hub_models.models.swin_small import Model
169
  torch_model = Model.from_pretrained()
170
 
171
  # Device
172
- device = hub.Device("Samsung Galaxy S24")
173
 
174
  # Trace model
175
  input_shape = torch_model.get_input_spec()
 
37
 
38
  | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
39
  |---|---|---|---|---|---|---|---|---|
40
+ | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 44.353 ms | 0 - 267 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
41
+ | Swin-Small | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 38.48 ms | 1 - 515 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
42
+ | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 23.466 ms | 0 - 261 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
43
+ | Swin-Small | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 24.133 ms | 1 - 233 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
44
+ | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 18.401 ms | 0 - 28 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
45
+ | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 15.814 ms | 0 - 59 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
46
+ | Swin-Small | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 15.449 ms | 0 - 54 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
47
+ | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 20.58 ms | 0 - 265 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
48
+ | Swin-Small | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 17.74 ms | 0 - 533 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
49
+ | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 44.353 ms | 0 - 267 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
50
+ | Swin-Small | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 38.48 ms | 1 - 515 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
51
+ | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 18.58 ms | 0 - 32 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
52
+ | Swin-Small | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 15.935 ms | 0 - 55 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
53
+ | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 26.558 ms | 0 - 258 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
54
+ | Swin-Small | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 23.452 ms | 1 - 509 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
55
+ | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 18.503 ms | 0 - 29 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
56
+ | Swin-Small | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 15.927 ms | 0 - 65 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
57
+ | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 20.58 ms | 0 - 265 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
58
+ | Swin-Small | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 17.74 ms | 0 - 533 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
59
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 12.565 ms | 0 - 273 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
60
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 10.323 ms | 1 - 783 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
61
+ | Swin-Small | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 10.069 ms | 1 - 732 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
62
+ | Swin-Small | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 10.164 ms | 0 - 266 MB | NPU | [Swin-Small.tflite](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.tflite) |
63
+ | Swin-Small | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 8.047 ms | 1 - 531 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
64
+ | Swin-Small | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 7.78 ms | 1 - 530 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
65
+ | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 16.638 ms | 596 - 596 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.dlc) |
66
+ | Swin-Small | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 16.067 ms | 101 - 101 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small.onnx.zip) |
67
+ | Swin-Small | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 28.291 ms | 0 - 302 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
68
+ | Swin-Small | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 19.659 ms | 0 - 228 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
69
+ | Swin-Small | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 15.498 ms | 0 - 65 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
70
+ | Swin-Small | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 126.654 ms | 270 - 431 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
71
+ | Swin-Small | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 15.963 ms | 0 - 303 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
72
+ | Swin-Small | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 49.55 ms | 0 - 737 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
73
+ | Swin-Small | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | ONNX | 341.927 ms | 66 - 93 MB | CPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
74
+ | Swin-Small | w8a16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 324.5 ms | 54 - 75 MB | CPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
75
+ | Swin-Small | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 28.291 ms | 0 - 302 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
76
+ | Swin-Small | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 15.479 ms | 0 - 72 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
77
+ | Swin-Small | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 18.856 ms | 0 - 303 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
78
+ | Swin-Small | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 15.495 ms | 0 - 76 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
79
+ | Swin-Small | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 15.963 ms | 0 - 303 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
80
+ | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 10.424 ms | 0 - 318 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
81
+ | Swin-Small | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 77.909 ms | 266 - 544 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
82
+ | Swin-Small | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 8.457 ms | 0 - 286 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
83
+ | Swin-Small | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 64.404 ms | 287 - 533 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
84
+ | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 16.402 ms | 241 - 241 MB | NPU | [Swin-Small.dlc](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.dlc) |
85
+ | Swin-Small | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 92.069 ms | 460 - 460 MB | NPU | [Swin-Small.onnx.zip](https://huggingface.co/qualcomm/Swin-Small/blob/main/Swin-Small_w8a16.onnx.zip) |
 
 
 
 
 
86
 
87
 
88
 
 
164
  torch_model = Model.from_pretrained()
165
 
166
  # Device
167
+ device = hub.Device("Samsung Galaxy S25")
168
 
169
  # Trace model
170
  input_shape = torch_model.get_input_spec()
Swin-Small_float.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e0b18e6e8dfc848bfd25dec281c4e01ad0f16a042aa283dad66e3797750c35cb
3
- size 202651052
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7af7749686ab3343c74d73a00aee602283ffc27ddaf5fe2bdd926dec24046cf1
3
+ size 202650548
Swin-Small_float.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5ea688a31a1eea5fd0ecab4a2b1db7bdbf896aba61775d9d429dd8c340a4eec5
3
- size 184744758
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1836d1c99aa8038f853b232d71704daab72e8feefee3bc1dc82684c36d67cedc
3
+ size 184744463
Swin-Small_w8a16.dlc CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:48ae5b80b0a3a9b7a018c137408be16f71c0289fa5d19b373e659d814a174a56
3
- size 55073588
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f8222b4dd5e07b22ddab63daa2db287b95ecf48d0fc0ec4a18f4a7503c1ca97
3
+ size 55073020
Swin-Small_w8a16.onnx.zip CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:58f463109a2c8df76c688752aec638cb019cc2ba270cf92c1786dd69c7a5ef43
3
- size 171031627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:954cad6a5f2c80bd43ffa7e0e24f5fe5e01a315bdd880fe7110b8489e2e58b97
3
+ size 171029884
precompiled/qualcomm-qcs6490-proxy/Swin-Small_w8a16.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:19e0d05f96f1b4f981b238cbcd639f3d9f2c4ef49dac10c6d9527d2b347f0dd6
3
- size 55996416
 
 
 
 
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/Swin-Small_float.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bbf7d721680ed0c80c65f7ba98624e6084b5018963023e7ae94e25d84e3b5622
3
- size 104960000
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/Swin-Small_float.onnx.zip DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a5933612b7be49cbdab7bc7938904ef31d5a0464b3e90c597f71ce7ed659c06e
3
- size 93595490
 
 
 
 
precompiled/qualcomm-snapdragon-x-elite/Swin-Small_w8a16.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a234e12bc2e30c7daee5bca980ccdd63a9837d8f9520b5563c4d49264c8505bd
3
- size 56397824
 
 
 
 
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