Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
@@ -1,10 +1,10 @@
|
|
1 |
---
|
2 |
library_name: pytorch
|
3 |
license: mit
|
4 |
-
pipeline_tag: automatic-speech-recognition
|
5 |
tags:
|
6 |
- foundation
|
7 |
- android
|
|
|
8 |
|
9 |
---
|
10 |
|
@@ -38,56 +38,64 @@ 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 |
-
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 38.
|
42 |
-
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 4.
|
43 |
-
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 16.
|
44 |
-
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE |
|
45 |
-
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 3.
|
46 |
-
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 13.
|
47 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE |
|
48 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 2.
|
49 |
-
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 12.
|
50 |
-
| WhisperDecoder |
|
51 |
-
| WhisperDecoder |
|
52 |
-
| WhisperDecoder |
|
53 |
-
| WhisperDecoder |
|
54 |
-
| WhisperDecoder |
|
55 |
-
| WhisperDecoder |
|
56 |
-
| WhisperDecoder |
|
57 |
-
| WhisperDecoder |
|
58 |
-
| WhisperDecoder |
|
59 |
-
| WhisperDecoder |
|
60 |
-
| WhisperDecoder |
|
61 |
-
| WhisperDecoder |
|
62 |
-
| WhisperDecoder |
|
63 |
-
| WhisperDecoder |
|
64 |
-
| WhisperDecoder |
|
65 |
-
| WhisperDecoder |
|
66 |
-
|
|
67 |
-
|
|
68 |
-
|
|
69 |
-
|
|
70 |
-
| WhisperEncoder | Samsung Galaxy
|
71 |
-
| WhisperEncoder | Samsung Galaxy
|
72 |
-
| WhisperEncoder |
|
73 |
-
| WhisperEncoder |
|
74 |
-
| WhisperEncoder |
|
75 |
-
| WhisperEncoder |
|
76 |
-
| WhisperEncoder |
|
77 |
-
| WhisperEncoder |
|
78 |
-
| WhisperEncoder |
|
79 |
-
| WhisperEncoder |
|
80 |
-
| WhisperEncoder |
|
81 |
-
| WhisperEncoder |
|
82 |
-
| WhisperEncoder |
|
83 |
-
| WhisperEncoder |
|
84 |
-
| WhisperEncoder |
|
85 |
-
| WhisperEncoder |
|
86 |
-
| WhisperEncoder |
|
87 |
-
| WhisperEncoder |
|
88 |
-
| WhisperEncoder |
|
89 |
-
| WhisperEncoder |
|
90 |
-
| WhisperEncoder |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
|
92 |
|
93 |
|
@@ -151,8 +159,8 @@ Profiling Results
|
|
151 |
WhisperDecoder
|
152 |
Device : Samsung Galaxy S23 (13)
|
153 |
Runtime : TFLITE
|
154 |
-
Estimated inference time (ms) : 38.
|
155 |
-
Estimated peak memory usage (MB): [
|
156 |
Total # Ops : 983
|
157 |
Compute Unit(s) : NPU (983 ops)
|
158 |
|
@@ -160,8 +168,8 @@ Compute Unit(s) : NPU (983 ops)
|
|
160 |
WhisperEncoder
|
161 |
Device : Samsung Galaxy S23 (13)
|
162 |
Runtime : TFLITE
|
163 |
-
Estimated inference time (ms) :
|
164 |
-
Estimated peak memory usage (MB): [0,
|
165 |
Total # Ops : 419
|
166 |
Compute Unit(s) : GPU (408 ops) CPU (11 ops)
|
167 |
```
|
|
|
1 |
---
|
2 |
library_name: pytorch
|
3 |
license: mit
|
|
|
4 |
tags:
|
5 |
- foundation
|
6 |
- android
|
7 |
+
pipeline_tag: automatic-speech-recognition
|
8 |
|
9 |
---
|
10 |
|
|
|
38 |
|
39 |
| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|
40 |
|---|---|---|---|---|---|---|---|---|
|
41 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 38.94 ms | 6 - 45 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
42 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 4.17 ms | 20 - 22 MB | FP16 | NPU | [Whisper-Base-En.so](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.so) |
|
43 |
+
| WhisperDecoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 16.458 ms | 11 - 305 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.onnx) |
|
44 |
+
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 32.578 ms | 0 - 81 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
45 |
+
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 3.235 ms | 7 - 23 MB | FP16 | NPU | [Whisper-Base-En.so](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.so) |
|
46 |
+
| WhisperDecoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 13.532 ms | 52 - 130 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.onnx) |
|
47 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 28.068 ms | 6 - 80 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
48 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 2.893 ms | 20 - 71 MB | FP16 | NPU | Use Export Script |
|
49 |
+
| WhisperDecoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 12.628 ms | 52 - 120 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.onnx) |
|
50 |
+
| WhisperDecoder | SA7255P ADP | SA7255P | TFLITE | 68.431 ms | 5 - 76 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
51 |
+
| WhisperDecoder | SA7255P ADP | SA7255P | QNN | 26.826 ms | 18 - 25 MB | FP16 | NPU | Use Export Script |
|
52 |
+
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 38.124 ms | 6 - 44 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
53 |
+
| WhisperDecoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 4.248 ms | 20 - 23 MB | FP16 | NPU | Use Export Script |
|
54 |
+
| WhisperDecoder | SA8295P ADP | SA8295P | TFLITE | 40.977 ms | 6 - 71 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
55 |
+
| WhisperDecoder | SA8295P ADP | SA8295P | QNN | 5.649 ms | 18 - 28 MB | FP16 | NPU | Use Export Script |
|
56 |
+
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 38.044 ms | 5 - 43 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
57 |
+
| WhisperDecoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 4.219 ms | 20 - 23 MB | FP16 | NPU | Use Export Script |
|
58 |
+
| WhisperDecoder | SA8775P ADP | SA8775P | TFLITE | 38.341 ms | 5 - 76 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
59 |
+
| WhisperDecoder | SA8775P ADP | SA8775P | QNN | 5.654 ms | 20 - 27 MB | FP16 | NPU | Use Export Script |
|
60 |
+
| WhisperDecoder | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 68.431 ms | 5 - 76 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
61 |
+
| WhisperDecoder | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 26.826 ms | 18 - 25 MB | FP16 | NPU | Use Export Script |
|
62 |
+
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 38.315 ms | 5 - 45 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
63 |
+
| WhisperDecoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 4.203 ms | 20 - 23 MB | FP16 | NPU | Use Export Script |
|
64 |
+
| WhisperDecoder | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 38.341 ms | 5 - 76 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
65 |
+
| WhisperDecoder | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 5.654 ms | 20 - 27 MB | FP16 | NPU | Use Export Script |
|
66 |
+
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 41.862 ms | 6 - 78 MB | FP16 | NPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.tflite) |
|
67 |
+
| WhisperDecoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 5.033 ms | 20 - 70 MB | FP16 | NPU | Use Export Script |
|
68 |
+
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 3.786 ms | 20 - 20 MB | FP16 | NPU | Use Export Script |
|
69 |
+
| WhisperDecoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 14.096 ms | 108 - 108 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperDecoder.onnx) |
|
70 |
+
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 201.775 ms | 0 - 76 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
71 |
+
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 216.587 ms | 1 - 2 MB | FP16 | NPU | [Whisper-Base-En.so](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.so) |
|
72 |
+
| WhisperEncoder | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 291.32 ms | 71 - 282 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.onnx) |
|
73 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 160.722 ms | 37 - 81 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
74 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 146.343 ms | 1 - 19 MB | FP16 | NPU | [Whisper-Base-En.so](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.so) |
|
75 |
+
| WhisperEncoder | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 204.336 ms | 116 - 448 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.onnx) |
|
76 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 158.856 ms | 39 - 65 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
77 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 151.385 ms | 0 - 304 MB | FP16 | NPU | Use Export Script |
|
78 |
+
| WhisperEncoder | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 176.092 ms | 42 - 374 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.onnx) |
|
79 |
+
| WhisperEncoder | SA7255P ADP | SA7255P | TFLITE | 1172.046 ms | 37 - 59 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
80 |
+
| WhisperEncoder | SA7255P ADP | SA7255P | QNN | 942.672 ms | 1 - 7 MB | FP16 | NPU | Use Export Script |
|
81 |
+
| WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 205.474 ms | 0 - 83 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
82 |
+
| WhisperEncoder | SA8255 (Proxy) | SA8255P Proxy | QNN | 219.13 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
|
83 |
+
| WhisperEncoder | SA8295P ADP | SA8295P | TFLITE | 209.558 ms | 38 - 62 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
84 |
+
| WhisperEncoder | SA8295P ADP | SA8295P | QNN | 218.817 ms | 1 - 11 MB | FP16 | NPU | Use Export Script |
|
85 |
+
| WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 245.615 ms | 0 - 55 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
86 |
+
| WhisperEncoder | SA8650 (Proxy) | SA8650P Proxy | QNN | 219.04 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
|
87 |
+
| WhisperEncoder | SA8775P ADP | SA8775P | TFLITE | 366.215 ms | 38 - 61 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
88 |
+
| WhisperEncoder | SA8775P ADP | SA8775P | QNN | 195.469 ms | 1 - 8 MB | FP16 | NPU | Use Export Script |
|
89 |
+
| WhisperEncoder | QCS8275 (Proxy) | QCS8275 Proxy | TFLITE | 1172.046 ms | 37 - 59 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
90 |
+
| WhisperEncoder | QCS8275 (Proxy) | QCS8275 Proxy | QNN | 942.672 ms | 1 - 7 MB | FP16 | NPU | Use Export Script |
|
91 |
+
| WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 208.371 ms | 0 - 70 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
92 |
+
| WhisperEncoder | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 214.059 ms | 1 - 3 MB | FP16 | NPU | Use Export Script |
|
93 |
+
| WhisperEncoder | QCS9075 (Proxy) | QCS9075 Proxy | TFLITE | 366.215 ms | 38 - 61 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
94 |
+
| WhisperEncoder | QCS9075 (Proxy) | QCS9075 Proxy | QNN | 195.469 ms | 1 - 8 MB | FP16 | NPU | Use Export Script |
|
95 |
+
| WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 279.395 ms | 38 - 87 MB | FP16 | GPU | [Whisper-Base-En.tflite](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.tflite) |
|
96 |
+
| WhisperEncoder | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 334.458 ms | 0 - 296 MB | FP16 | NPU | Use Export Script |
|
97 |
+
| WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 160.586 ms | 0 - 0 MB | FP16 | NPU | Use Export Script |
|
98 |
+
| WhisperEncoder | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 296.449 ms | 133 - 133 MB | FP16 | NPU | [Whisper-Base-En.onnx](https://huggingface.co/qualcomm/Whisper-Base-En/blob/main/WhisperEncoder.onnx) |
|
99 |
|
100 |
|
101 |
|
|
|
159 |
WhisperDecoder
|
160 |
Device : Samsung Galaxy S23 (13)
|
161 |
Runtime : TFLITE
|
162 |
+
Estimated inference time (ms) : 38.9
|
163 |
+
Estimated peak memory usage (MB): [6, 45]
|
164 |
Total # Ops : 983
|
165 |
Compute Unit(s) : NPU (983 ops)
|
166 |
|
|
|
168 |
WhisperEncoder
|
169 |
Device : Samsung Galaxy S23 (13)
|
170 |
Runtime : TFLITE
|
171 |
+
Estimated inference time (ms) : 201.8
|
172 |
+
Estimated peak memory usage (MB): [0, 76]
|
173 |
Total # Ops : 419
|
174 |
Compute Unit(s) : GPU (408 ops) CPU (11 ops)
|
175 |
```
|