Hugging Face's logo

qualcomm
/
ConvNext-Tiny

Image Classification
PyTorch
LiteRT
ONNX
android
Model card Files Community
bhushans commited on
Commit
62e501f
·
verified ·
1 Parent(s): 2369039

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +44 -26
README.md CHANGED
@@ -34,29 +34,31 @@ More details on model performance across various devices, can be found
34
 
35
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  |---|---|---|---|---|---|---|---|---|
37
- | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 3.36 ms | 0 - 2 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
38
- | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 3.919 ms | 1 - 189 MB | FP16 | NPU | [ConvNext-Tiny.so](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.so) |
39
- | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 13.496 ms | 0 - 65 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
40
- | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 2.459 ms | 0 - 212 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
41
- | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 2.835 ms | 0 - 35 MB | FP16 | NPU | [ConvNext-Tiny.so](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.so) |
42
- | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 9.529 ms | 1 - 371 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
43
- | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 2.141 ms | 0 - 61 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
44
- | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 2.437 ms | 1 - 36 MB | FP16 | NPU | Use Export Script |
45
- | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 8.676 ms | 0 - 125 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
46
- | ConvNext-Tiny | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 3.36 ms | 0 - 2 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
47
- | ConvNext-Tiny | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 3.624 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
48
- | ConvNext-Tiny | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 3.396 ms | 0 - 2 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
49
- | ConvNext-Tiny | SA8255 (Proxy) | SA8255P Proxy | QNN | 3.703 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
50
- | ConvNext-Tiny | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 3.361 ms | 0 - 2 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
51
- | ConvNext-Tiny | SA8775 (Proxy) | SA8775P Proxy | QNN | 3.673 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
52
- | ConvNext-Tiny | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 3.352 ms | 0 - 19 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
53
- | ConvNext-Tiny | SA8650 (Proxy) | SA8650P Proxy | QNN | 3.631 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
54
- | ConvNext-Tiny | SA8295P ADP | SA8295P | TFLITE | 10.471 ms | 0 - 53 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
55
- | ConvNext-Tiny | SA8295P ADP | SA8295P | QNN | 9.481 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
56
- | ConvNext-Tiny | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 9.193 ms | 0 - 199 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
57
- | ConvNext-Tiny | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 9.845 ms | 0 - 34 MB | FP16 | NPU | Use Export Script |
58
- | ConvNext-Tiny | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.898 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
59
- | ConvNext-Tiny | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 16.269 ms | 58 - 58 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
 
 
60
 
61
 
62
 
@@ -120,8 +122,8 @@ Profiling Results
120
  ConvNext-Tiny
121
  Device : Samsung Galaxy S23 (13)
122
  Runtime : TFLITE
123
- Estimated inference time (ms) : 3.4
124
- Estimated peak memory usage (MB): [0, 2]
125
  Total # Ops : 328
126
  Compute Unit(s) : NPU (328 ops)
127
  ```
@@ -142,13 +144,29 @@ in memory using the `jit.trace` and then call the `submit_compile_job` API.
142
  import torch
143
 
144
  import qai_hub as hub
145
- from qai_hub_models.models.convnext_tiny import
146
 
147
  # Load the model
 
148
 
149
  # Device
150
  device = hub.Device("Samsung Galaxy S23")
151
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
152
 
153
  ```
154
 
 
34
 
35
  | Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
36
  |---|---|---|---|---|---|---|---|---|
37
+ | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 3.346 ms | 0 - 259 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
38
+ | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 3.949 ms | 1 - 279 MB | FP16 | NPU | [ConvNext-Tiny.so](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.so) |
39
+ | ConvNext-Tiny | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 13.377 ms | 0 - 65 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
40
+ | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 2.447 ms | 0 - 35 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
41
+ | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 2.82 ms | 1 - 36 MB | FP16 | NPU | [ConvNext-Tiny.so](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.so) |
42
+ | ConvNext-Tiny | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 9.597 ms | 0 - 370 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
43
+ | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 1.846 ms | 0 - 37 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
44
+ | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 2.445 ms | 0 - 37 MB | FP16 | NPU | Use Export Script |
45
+ | ConvNext-Tiny | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 7.291 ms | 1 - 127 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
46
+ | ConvNext-Tiny | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 3.342 ms | 0 - 319 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
47
+ | ConvNext-Tiny | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 3.641 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
48
+ | ConvNext-Tiny | SA7255P ADP | SA7255P | TFLITE | 96.726 ms | 0 - 35 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
49
+ | ConvNext-Tiny | SA7255P ADP | SA7255P | QNN | 97.255 ms | 1 - 8 MB | FP16 | NPU | Use Export Script |
50
+ | ConvNext-Tiny | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 3.344 ms | 0 - 269 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
51
+ | ConvNext-Tiny | SA8255 (Proxy) | SA8255P Proxy | QNN | 3.716 ms | 0 - 1 MB | FP16 | NPU | Use Export Script |
52
+ | ConvNext-Tiny | SA8295P ADP | SA8295P | TFLITE | 11.15 ms | 0 - 32 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
53
+ | ConvNext-Tiny | SA8295P ADP | SA8295P | QNN | 9.48 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
54
+ | ConvNext-Tiny | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 3.374 ms | 0 - 289 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
55
+ | ConvNext-Tiny | SA8650 (Proxy) | SA8650P Proxy | QNN | 3.715 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
56
+ | ConvNext-Tiny | SA8775P ADP | SA8775P | TFLITE | 5.694 ms | 0 - 35 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
57
+ | ConvNext-Tiny | SA8775P ADP | SA8775P | QNN | 6.182 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
58
+ | ConvNext-Tiny | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 9.806 ms | 0 - 34 MB | FP16 | NPU | [ConvNext-Tiny.tflite](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.tflite) |
59
+ | ConvNext-Tiny | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 10.415 ms | 0 - 34 MB | FP16 | NPU | Use Export Script |
60
+ | ConvNext-Tiny | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.888 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
61
+ | ConvNext-Tiny | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 16.248 ms | 57 - 57 MB | FP16 | NPU | [ConvNext-Tiny.onnx](https://huggingface.co/qualcomm/ConvNext-Tiny/blob/main/ConvNext-Tiny.onnx) |
62
 
63
 
64
 
 
122
  ConvNext-Tiny
123
  Device : Samsung Galaxy S23 (13)
124
  Runtime : TFLITE
125
+ Estimated inference time (ms) : 3.3
126
+ Estimated peak memory usage (MB): [0, 259]
127
  Total # Ops : 328
128
  Compute Unit(s) : NPU (328 ops)
129
  ```
 
144
  import torch
145
 
146
  import qai_hub as hub
147
+ from qai_hub_models.models.convnext_tiny import Model
148
 
149
  # Load the model
150
+ torch_model = Model.from_pretrained()
151
 
152
  # Device
153
  device = hub.Device("Samsung Galaxy S23")
154
 
155
+ # Trace model
156
+ input_shape = torch_model.get_input_spec()
157
+ sample_inputs = torch_model.sample_inputs()
158
+
159
+ pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
160
+
161
+ # Compile model on a specific device
162
+ compile_job = hub.submit_compile_job(
163
+ model=pt_model,
164
+ device=device,
165
+ input_specs=torch_model.get_input_spec(),
166
+ )
167
+
168
+ # Get target model to run on-device
169
+ target_model = compile_job.get_target_model()
170
 
171
  ```
172