Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
@@ -38,49 +38,49 @@ More details on model performance across various devices, can be found
|
|
38 |
|
39 |
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
40 |
|---|---|---|---|---|---|---|---|---|
|
41 |
-
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE |
|
42 |
-
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN |
|
43 |
-
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE |
|
44 |
-
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 |
|
45 |
-
| WhisperEncoder |
|
46 |
-
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite |
|
47 |
-
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite |
|
48 |
-
| WhisperEncoder |
|
49 |
-
| WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy |
|
50 |
-
| WhisperEncoder |
|
51 |
-
| WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy |
|
52 |
-
| WhisperEncoder |
|
53 |
-
| WhisperEncoder | SA8775 (Proxy) | SA8775P Proxy |
|
54 |
-
| WhisperEncoder |
|
55 |
-
| WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy |
|
56 |
-
| WhisperEncoder |
|
57 |
-
| WhisperEncoder | SA8295P ADP | SA8295P |
|
58 |
-
| WhisperEncoder |
|
59 |
-
| WhisperEncoder |
|
60 |
-
| WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite |
|
61 |
-
|
|
62 |
-
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 |
|
63 |
-
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 |
|
64 |
-
| WhisperDecoder | Samsung Galaxy
|
65 |
-
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 |
|
66 |
-
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 |
|
67 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE |
|
68 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN |
|
69 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 39.
|
70 |
-
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE |
|
71 |
-
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 12.
|
72 |
-
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 25.
|
73 |
-
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 12.
|
74 |
-
| WhisperDecoder | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 25.
|
75 |
-
| WhisperDecoder | SA8775 (Proxy) | SA8775P Proxy | QNN | 12.
|
76 |
-
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE |
|
77 |
-
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 12.
|
78 |
-
| WhisperDecoder | SA8295P ADP | SA8295P | TFLITE | 27.
|
79 |
-
| WhisperDecoder | SA8295P ADP | SA8295P | QNN | 14.
|
80 |
-
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE |
|
81 |
-
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN |
|
82 |
-
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 10.
|
83 |
-
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX |
|
84 |
|
85 |
|
86 |
|
@@ -145,8 +145,8 @@ Profiling Results
|
|
145 |
WhisperEncoder
|
146 |
Device : Samsung Galaxy S23 (13)
|
147 |
Runtime : TFLITE
|
148 |
-
Estimated inference time (ms) :
|
149 |
-
Estimated peak memory usage (MB): [
|
150 |
Total # Ops : 911
|
151 |
Compute Unit(s) : GPU (900 ops) CPU (11 ops)
|
152 |
|
@@ -154,7 +154,7 @@ Compute Unit(s) : GPU (900 ops) CPU (11 ops)
|
|
154 |
WhisperDecoder
|
155 |
Device : Samsung Galaxy S23 (13)
|
156 |
Runtime : TFLITE
|
157 |
-
Estimated inference time (ms) : 25.
|
158 |
Estimated peak memory usage (MB): [16, 19]
|
159 |
Total # Ops : 2573
|
160 |
Compute Unit(s) : NPU (2573 ops)
|
|
|
38 |
|
39 |
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
40 |
|---|---|---|---|---|---|---|---|---|
|
41 |
+
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 704.009 ms | 45 - 439 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
42 |
+
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 866.62 ms | 2 - 230 MB | FP16 | NPU | [Whisper-Small-En.so](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.so) |
|
43 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 514.369 ms | 108 - 196 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
44 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 615.589 ms | 0 - 839 MB | FP16 | NPU | [Whisper-Small-En.so](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.so) |
|
45 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 808.662 ms | 145 - 4336 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.onnx) |
|
46 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 537.047 ms | 111 - 139 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
47 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 494.644 ms | 0 - 908 MB | FP16 | NPU | Use Export Script |
|
48 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 698.505 ms | 117 - 2773 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.onnx) |
|
49 |
+
| WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 686.962 ms | 40 - 440 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
50 |
+
| WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 693.295 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
|
51 |
+
| WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 711.433 ms | 18 - 355 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
52 |
+
| WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 701.7 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
|
53 |
+
| WhisperEncoder | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 703.544 ms | 61 - 457 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
54 |
+
| WhisperEncoder | SA8775 (Proxy) | SA8775P Proxy | QNN | 713.618 ms | 0 - 57 MB | FP16 | NPU | Use Export Script |
|
55 |
+
| WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 712.376 ms | 33 - 421 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
56 |
+
| WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 724.601 ms | 1 - 30 MB | FP16 | NPU | Use Export Script |
|
57 |
+
| WhisperEncoder | SA8295P ADP | SA8295P | TFLITE | 658.817 ms | 108 - 140 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
58 |
+
| WhisperEncoder | SA8295P ADP | SA8295P | QNN | 727.239 ms | 3 - 8 MB | FP16 | NPU | Use Export Script |
|
59 |
+
| WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 972.067 ms | 75 - 170 MB | FP16 | GPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.tflite) |
|
60 |
+
| WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 526.155 ms | 0 - 0 MB | FP16 | NPU | Use Export Script |
|
61 |
+
| WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1356.64 ms | 449 - 449 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperEncoder.onnx) |
|
62 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 25.328 ms | 16 - 19 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
63 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 12.008 ms | 61 - 130 MB | FP16 | NPU | [Whisper-Small-En.so](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.so) |
|
64 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 57.142 ms | 121 - 124 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.onnx) |
|
65 |
+
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 19.228 ms | 32 - 1144 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
66 |
+
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 9.452 ms | 59 - 158 MB | FP16 | NPU | [Whisper-Small-En.so](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.so) |
|
67 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 16.628 ms | 16 - 263 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
68 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 7.482 ms | 50 - 183 MB | FP16 | NPU | Use Export Script |
|
69 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 39.601 ms | 111 - 883 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.onnx) |
|
70 |
+
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 25.84 ms | 16 - 18 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
71 |
+
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 12.335 ms | 57 - 58 MB | FP16 | NPU | Use Export Script |
|
72 |
+
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 25.356 ms | 15 - 18 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
73 |
+
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 12.62 ms | 64 - 65 MB | FP16 | NPU | Use Export Script |
|
74 |
+
| WhisperDecoder | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 25.355 ms | 14 - 18 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
75 |
+
| WhisperDecoder | SA8775 (Proxy) | SA8775P Proxy | QNN | 12.69 ms | 65 - 66 MB | FP16 | NPU | Use Export Script |
|
76 |
+
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 24.58 ms | 15 - 18 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
77 |
+
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 12.971 ms | 61 - 62 MB | FP16 | NPU | Use Export Script |
|
78 |
+
| WhisperDecoder | SA8295P ADP | SA8295P | TFLITE | 27.039 ms | 16 - 243 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
79 |
+
| WhisperDecoder | SA8295P ADP | SA8295P | QNN | 14.311 ms | 57 - 62 MB | FP16 | NPU | Use Export Script |
|
80 |
+
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 28.297 ms | 16 - 1104 MB | FP16 | NPU | [Whisper-Small-En.tflite](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.tflite) |
|
81 |
+
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 15.403 ms | 57 - 156 MB | FP16 | NPU | Use Export Script |
|
82 |
+
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 10.907 ms | 61 - 61 MB | FP16 | NPU | Use Export Script |
|
83 |
+
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 49.551 ms | 232 - 232 MB | FP16 | NPU | [Whisper-Small-En.onnx](https://huggingface.co/qualcomm/Whisper-Small-En/blob/main/WhisperDecoder.onnx) |
|
84 |
|
85 |
|
86 |
|
|
|
145 |
WhisperEncoder
|
146 |
Device : Samsung Galaxy S23 (13)
|
147 |
Runtime : TFLITE
|
148 |
+
Estimated inference time (ms) : 704.0
|
149 |
+
Estimated peak memory usage (MB): [45, 439]
|
150 |
Total # Ops : 911
|
151 |
Compute Unit(s) : GPU (900 ops) CPU (11 ops)
|
152 |
|
|
|
154 |
WhisperDecoder
|
155 |
Device : Samsung Galaxy S23 (13)
|
156 |
Runtime : TFLITE
|
157 |
+
Estimated inference time (ms) : 25.3
|
158 |
Estimated peak memory usage (MB): [16, 19]
|
159 |
Total # Ops : 2573
|
160 |
Compute Unit(s) : NPU (2573 ops)
|