Hugging Face's logo

qualcomm
/
WideResNet50-Quantized

Image Classification
PyTorch
TF Lite
ONNX
backbone
quantized
android
Model card Files Community
qaihm-bot commited on
Commit
9a5982f
·
verified ·
1 Parent(s): 9dc964e

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +33 -32
README.md CHANGED
@@ -36,42 +36,42 @@ More details on model performance across various devices, can be found
36
 
37
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
38
  |---|---|---|---|---|---|---|---|---|
39
- | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 1.781 ms | 0 - 140 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
40
- | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.034 ms | 0 - 139 MB | INT8 | NPU | [WideResNet50-Quantized.so](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.so) |
41
- | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 2.435 ms | 0 - 162 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
42
- | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 1.353 ms | 0 - 31 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
43
- | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.547 ms | 0 - 27 MB | INT8 | NPU | [WideResNet50-Quantized.so](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.so) |
44
- | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 1.778 ms | 0 - 33 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
45
- | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.249 ms | 0 - 25 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
46
- | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.459 ms | 0 - 22 MB | INT8 | NPU | Use Export Script |
47
- | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 1.707 ms | 0 - 33 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
48
- | WideResNet50-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 7.98 ms | 1 - 27 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
49
- | WideResNet50-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | QNN | 9.291 ms | 0 - 11 MB | INT8 | NPU | Use Export Script |
50
- | WideResNet50-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 24.129 ms | 0 - 2 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
51
- | WideResNet50-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 1.775 ms | 0 - 120 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
52
- | WideResNet50-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 1.922 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
53
- | WideResNet50-Quantized | SA7255P ADP | SA7255P | TFLITE | 22.358 ms | 0 - 18 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
54
- | WideResNet50-Quantized | SA7255P ADP | SA7255P | QNN | 23.727 ms | 0 - 10 MB | INT8 | NPU | Use Export Script |
55
- | WideResNet50-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 1.774 ms | 0 - 129 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
56
- | WideResNet50-Quantized | SA8255 (Proxy) | SA8255P Proxy | QNN | 1.925 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
57
- | WideResNet50-Quantized | SA8295P ADP | SA8295P | TFLITE | 2.732 ms | 0 - 26 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
58
- | WideResNet50-Quantized | SA8295P ADP | SA8295P | QNN | 3.06 ms | 0 - 14 MB | INT8 | NPU | Use Export Script |
59
- | WideResNet50-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 1.787 ms | 0 - 120 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
60
- | WideResNet50-Quantized | SA8650 (Proxy) | SA8650P Proxy | QNN | 1.97 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
61
- | WideResNet50-Quantized | SA8775P ADP | SA8775P | TFLITE | 2.071 ms | 0 - 18 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
62
- | WideResNet50-Quantized | SA8775P ADP | SA8775P | QNN | 2.463 ms | 0 - 10 MB | INT8 | NPU | Use Export Script |
63
- | WideResNet50-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 2.228 ms | 0 - 31 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
64
- | WideResNet50-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.481 ms | 0 - 25 MB | INT8 | NPU | Use Export Script |
65
- | WideResNet50-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 1.846 ms | 0 - 0 MB | INT8 | NPU | Use Export Script |
66
- | WideResNet50-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.472 ms | 70 - 70 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
67
 
68
 
69
 
70
 
71
  ## Installation
72
 
73
- This model can be installed as a Python package via pip.
74
 
 
75
  ```bash
76
  pip install qai-hub-models
77
  ```
@@ -128,7 +128,7 @@ WideResNet50-Quantized
128
  Device : Samsung Galaxy S23 (13)
129
  Runtime : TFLITE
130
  Estimated inference time (ms) : 1.8
131
- Estimated peak memory usage (MB): [0, 140]
132
  Total # Ops : 82
133
  Compute Unit(s) : NPU (82 ops)
134
  ```
@@ -155,7 +155,7 @@ from qai_hub_models.models.wideresnet50_quantized import Model
155
  torch_model = Model.from_pretrained()
156
 
157
  # Device
158
- device = hub.Device("Samsung Galaxy S23")
159
 
160
  # Trace model
161
  input_shape = torch_model.get_input_spec()
@@ -247,7 +247,8 @@ Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
247
 
248
 
249
  ## License
250
- * The license for the original implementation of WideResNet50-Quantized can be found [here](https://github.com/pytorch/vision/blob/main/LICENSE).
 
251
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
252
 
253
 
 
36
 
37
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
38
  |---|---|---|---|---|---|---|---|---|
39
+ | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 1.783 ms | 0 - 130 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
40
+ | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 2.044 ms | 0 - 139 MB | INT8 | NPU | [WideResNet50-Quantized.so](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.so) |
41
+ | WideResNet50-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 2.397 ms | 0 - 85 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
42
+ | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 1.362 ms | 0 - 32 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
43
+ | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 1.536 ms | 0 - 25 MB | INT8 | NPU | [WideResNet50-Quantized.so](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.so) |
44
+ | WideResNet50-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 1.763 ms | 0 - 35 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
45
+ | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.216 ms | 0 - 26 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
46
+ | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 1.335 ms | 0 - 26 MB | INT8 | NPU | Use Export Script |
47
+ | WideResNet50-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 1.737 ms | 0 - 33 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
48
+ | WideResNet50-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 8.225 ms | 0 - 26 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
49
+ | WideResNet50-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | QNN | 9.312 ms | 0 - 12 MB | INT8 | NPU | Use Export Script |
50
+ | WideResNet50-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 24.459 ms | 0 - 3 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
51
+ | WideResNet50-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 1.779 ms | 0 - 131 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
52
+ | WideResNet50-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 1.92 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
53
+ | WideResNet50-Quantized | SA7255P ADP | SA7255P | TFLITE | 22.239 ms | 0 - 18 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
54
+ | WideResNet50-Quantized | SA7255P ADP | SA7255P | QNN | 23.544 ms | 0 - 10 MB | INT8 | NPU | Use Export Script |
55
+ | WideResNet50-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 1.778 ms | 0 - 9 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
56
+ | WideResNet50-Quantized | SA8255 (Proxy) | SA8255P Proxy | QNN | 1.919 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
57
+ | WideResNet50-Quantized | SA8295P ADP | SA8295P | TFLITE | 2.718 ms | 0 - 26 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
58
+ | WideResNet50-Quantized | SA8295P ADP | SA8295P | QNN | 2.92 ms | 0 - 14 MB | INT8 | NPU | Use Export Script |
59
+ | WideResNet50-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 1.778 ms | 0 - 130 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
60
+ | WideResNet50-Quantized | SA8650 (Proxy) | SA8650P Proxy | QNN | 1.927 ms | 0 - 3 MB | INT8 | NPU | Use Export Script |
61
+ | WideResNet50-Quantized | SA8775P ADP | SA8775P | TFLITE | 2.085 ms | 0 - 18 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
62
+ | WideResNet50-Quantized | SA8775P ADP | SA8775P | QNN | 2.455 ms | 0 - 10 MB | INT8 | NPU | Use Export Script |
63
+ | WideResNet50-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 2.191 ms | 0 - 32 MB | INT8 | NPU | [WideResNet50-Quantized.tflite](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.tflite) |
64
+ | WideResNet50-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 2.575 ms | 0 - 28 MB | INT8 | NPU | Use Export Script |
65
+ | WideResNet50-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 1.824 ms | 0 - 0 MB | INT8 | NPU | Use Export Script |
66
+ | WideResNet50-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.494 ms | 70 - 70 MB | INT8 | NPU | [WideResNet50-Quantized.onnx](https://huggingface.co/qualcomm/WideResNet50-Quantized/blob/main/WideResNet50-Quantized.onnx) |
67
 
68
 
69
 
70
 
71
  ## Installation
72
 
 
73
 
74
+ Install the package via pip:
75
  ```bash
76
  pip install qai-hub-models
77
  ```
 
128
  Device : Samsung Galaxy S23 (13)
129
  Runtime : TFLITE
130
  Estimated inference time (ms) : 1.8
131
+ Estimated peak memory usage (MB): [0, 130]
132
  Total # Ops : 82
133
  Compute Unit(s) : NPU (82 ops)
134
  ```
 
155
  torch_model = Model.from_pretrained()
156
 
157
  # Device
158
+ device = hub.Device("Samsung Galaxy S24")
159
 
160
  # Trace model
161
  input_shape = torch_model.get_input_spec()
 
247
 
248
 
249
  ## License
250
+ * The license for the original implementation of WideResNet50-Quantized can be found
251
+ [here](https://github.com/pytorch/vision/blob/main/LICENSE).
252
  * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
253
 
254