Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
@@ -37,24 +37,24 @@ More details on model performance across various devices, can be found
|
|
37 |
|
38 |
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
39 |
|---|---|---|---|---|---|---|---|---|
|
40 |
-
| FFNet-40S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 8.
|
41 |
-
| FFNet-40S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX |
|
42 |
-
| FFNet-40S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 6.
|
43 |
-
| FFNet-40S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX |
|
44 |
-
| FFNet-40S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 6.
|
45 |
-
| FFNet-40S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX |
|
46 |
-
| FFNet-40S-Quantized | SA7255P ADP | SA7255P | TFLITE | 105.
|
47 |
-
| FFNet-40S-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 8.
|
48 |
-
| FFNet-40S-Quantized | SA8295P ADP | SA8295P | TFLITE | 13.
|
49 |
-
| FFNet-40S-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE |
|
50 |
-
| FFNet-40S-Quantized | SA8775P ADP | SA8775P | TFLITE | 11.
|
51 |
-
| FFNet-40S-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 56.
|
52 |
-
| FFNet-40S-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE |
|
53 |
-
| FFNet-40S-Quantized | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 105.
|
54 |
-
| FFNet-40S-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 8.
|
55 |
-
| FFNet-40S-Quantized | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 11.
|
56 |
-
| FFNet-40S-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 10.
|
57 |
-
| FFNet-40S-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX |
|
58 |
|
59 |
|
60 |
|
@@ -119,7 +119,7 @@ FFNet-40S-Quantized
|
|
119 |
Device : Samsung Galaxy S23 (13)
|
120 |
Runtime : TFLITE
|
121 |
Estimated inference time (ms) : 9.0
|
122 |
-
Estimated peak memory usage (MB): [
|
123 |
Total # Ops : 96
|
124 |
Compute Unit(s) : NPU (96 ops)
|
125 |
```
|
|
|
37 |
|
38 |
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
39 |
|---|---|---|---|---|---|---|---|---|
|
40 |
+
| FFNet-40S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 8.956 ms | 1 - 17 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
41 |
+
| FFNet-40S-Quantized | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 16.946 ms | 1 - 40 MB | INT8 | NPU | [FFNet-40S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.onnx) |
|
42 |
+
| FFNet-40S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 6.463 ms | 0 - 50 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
43 |
+
| FFNet-40S-Quantized | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 12.455 ms | 10 - 96 MB | INT8 | NPU | [FFNet-40S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.onnx) |
|
44 |
+
| FFNet-40S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 6.365 ms | 0 - 37 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
45 |
+
| FFNet-40S-Quantized | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 13.132 ms | 0 - 80 MB | INT8 | NPU | [FFNet-40S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.onnx) |
|
46 |
+
| FFNet-40S-Quantized | SA7255P ADP | SA7255P | TFLITE | 105.457 ms | 0 - 33 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
47 |
+
| FFNet-40S-Quantized | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 8.946 ms | 1 - 16 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
48 |
+
| FFNet-40S-Quantized | SA8295P ADP | SA8295P | TFLITE | 13.588 ms | 1 - 36 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
49 |
+
| FFNet-40S-Quantized | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 8.847 ms | 1 - 16 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
50 |
+
| FFNet-40S-Quantized | SA8775P ADP | SA8775P | TFLITE | 11.708 ms | 1 - 34 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
51 |
+
| FFNet-40S-Quantized | RB3 Gen 2 (Proxy) | QCS6490 Proxy | TFLITE | 56.087 ms | 1 - 53 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
52 |
+
| FFNet-40S-Quantized | RB5 (Proxy) | QCS8250 Proxy | TFLITE | 306.708 ms | 1 - 3 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
53 |
+
| FFNet-40S-Quantized | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 105.457 ms | 0 - 33 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
54 |
+
| FFNet-40S-Quantized | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 8.816 ms | 1 - 10 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
55 |
+
| FFNet-40S-Quantized | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 11.708 ms | 1 - 34 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
56 |
+
| FFNet-40S-Quantized | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 10.434 ms | 0 - 56 MB | INT8 | NPU | [FFNet-40S-Quantized.tflite](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.tflite) |
|
57 |
+
| FFNet-40S-Quantized | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 17.532 ms | 8 - 8 MB | INT8 | NPU | [FFNet-40S-Quantized.onnx](https://huggingface.co/qualcomm/FFNet-40S-Quantized/blob/main/FFNet-40S-Quantized.onnx) |
|
58 |
|
59 |
|
60 |
|
|
|
119 |
Device : Samsung Galaxy S23 (13)
|
120 |
Runtime : TFLITE
|
121 |
Estimated inference time (ms) : 9.0
|
122 |
+
Estimated peak memory usage (MB): [1, 17]
|
123 |
Total # Ops : 96
|
124 |
Compute Unit(s) : NPU (96 ops)
|
125 |
```
|