v0.63.0
Browse filesSee https://github.com/qualcomm/ai-hub-models/releases/v0.63.0 for changelog.
- README.md +41 -45
- release_assets.json +7 -7
README.md
CHANGED
|
@@ -14,7 +14,7 @@ pipeline_tag: robotics
|
|
| 14 |
ACT (Action Chunking with Transformers) is a robotic policy model that is trained to predict the next chunk of actions that the robotic hand is expected to perform.
|
| 15 |
|
| 16 |
This is based on the implementation of ACT found [here](https://github.com/tonyzhaozh/act).
|
| 17 |
-
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.
|
| 18 |
|
| 19 |
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
|
| 20 |
|
|
@@ -27,23 +27,23 @@ Below are pre-exported model assets ready for deployment.
|
|
| 27 |
|
| 28 |
| Runtime | Precision | Chipset | SDK Versions | Download |
|
| 29 |
|---|---|---|---|---|
|
| 30 |
-
| ONNX | float | Universal | QAIRT 2.
|
| 31 |
-
| QNN_DLC | float | Universal | QAIRT 2.
|
| 32 |
-
| TFLITE | float | Universal | QAIRT 2.
|
| 33 |
|
| 34 |
For more device-specific assets and performance metrics, visit **[ACT on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/act)**.
|
| 35 |
|
| 36 |
|
| 37 |
### Option 2: Export with Custom Configurations
|
| 38 |
|
| 39 |
-
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.
|
| 40 |
- Custom weights (e.g., fine-tuned checkpoints)
|
| 41 |
- Custom input shapes
|
| 42 |
- Target device and runtime configurations
|
| 43 |
|
| 44 |
This option is ideal if you need to customize the model beyond the default configuration provided here.
|
| 45 |
|
| 46 |
-
See our repository for [ACT on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.
|
| 47 |
|
| 48 |
## Model Details
|
| 49 |
|
|
@@ -58,49 +58,45 @@ See our repository for [ACT on GitHub](https://github.com/qualcomm/ai-hub-models
|
|
| 58 |
## Performance Summary
|
| 59 |
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|
| 60 |
|---|---|---|---|---|---|---
|
| 61 |
-
| ACT | ONNX | float | Snapdragon®
|
| 62 |
-
| ACT | ONNX | float | Snapdragon®
|
| 63 |
-
| ACT | ONNX | float | Snapdragon®
|
| 64 |
-
| ACT | ONNX | float | Snapdragon®
|
| 65 |
-
| ACT | ONNX | float |
|
| 66 |
-
| ACT | ONNX | float |
|
| 67 |
-
| ACT | ONNX | float | Qualcomm®
|
| 68 |
-
| ACT | ONNX | float | Qualcomm® Dragonwing™
|
| 69 |
-
| ACT | ONNX | float | Qualcomm®
|
| 70 |
-
| ACT | ONNX | float | Qualcomm® Dragonwing™
|
| 71 |
-
| ACT | ONNX | float |
|
| 72 |
-
| ACT | ONNX | float |
|
| 73 |
-
| ACT | QNN_DLC | float | Snapdragon®
|
| 74 |
-
| ACT | QNN_DLC | float | Snapdragon®
|
| 75 |
-
| ACT | QNN_DLC | float | Snapdragon®
|
| 76 |
-
| ACT | QNN_DLC | float | Snapdragon®
|
| 77 |
-
| ACT | QNN_DLC | float |
|
| 78 |
-
| ACT | QNN_DLC | float |
|
| 79 |
-
| ACT | QNN_DLC | float | Qualcomm®
|
|
|
|
| 80 |
| ACT | QNN_DLC | float | Qualcomm® SA8650P | 14.499 ms | 1 - 271 MB | NPU
|
| 81 |
| ACT | QNN_DLC | float | Qualcomm® SA8255P | 14.499 ms | 1 - 271 MB | NPU
|
| 82 |
-
| ACT | QNN_DLC | float | Qualcomm® QCS8450 |
|
| 83 |
-
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 14.
|
| 84 |
-
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 |
|
| 85 |
-
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 5.
|
| 86 |
-
| ACT | QNN_DLC | float | Qualcomm®
|
| 87 |
-
| ACT |
|
| 88 |
-
| ACT |
|
| 89 |
-
| ACT |
|
| 90 |
-
| ACT | TFLITE | float | Snapdragon® 8 Gen
|
| 91 |
-
| ACT | TFLITE | float |
|
| 92 |
-
| ACT | TFLITE | float | Qualcomm® Dragonwing™
|
| 93 |
-
| ACT | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.509 ms | 0 - 3 MB | NPU
|
| 94 |
-
| ACT | TFLITE | float | Qualcomm® SA8775P | 13.799 ms | 0 - 279 MB | NPU
|
| 95 |
| ACT | TFLITE | float | Qualcomm® SA8650P | 13.799 ms | 0 - 279 MB | NPU
|
| 96 |
| ACT | TFLITE | float | Qualcomm® SA8255P | 13.799 ms | 0 - 279 MB | NPU
|
| 97 |
-
| ACT | TFLITE | float | Qualcomm® QCS8450 |
|
| 98 |
-
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 |
|
| 99 |
-
| ACT | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.
|
| 100 |
-
| ACT | TFLITE | float | Qualcomm®
|
| 101 |
-
| ACT | TFLITE | float | Qualcomm® SA8295P | 15.716 ms | 0 - 207 MB | NPU
|
| 102 |
-
| ACT | TFLITE | float | Snapdragon® 8 Elite Mobile | 4.738 ms | 0 - 282 MB | NPU
|
| 103 |
-
| ACT | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.794 ms | 0 - 324 MB | NPU
|
| 104 |
|
| 105 |
## License
|
| 106 |
* The license for the original implementation of ACT can be found
|
|
|
|
| 14 |
ACT (Action Chunking with Transformers) is a robotic policy model that is trained to predict the next chunk of actions that the robotic hand is expected to perform.
|
| 15 |
|
| 16 |
This is based on the implementation of ACT found [here](https://github.com/tonyzhaozh/act).
|
| 17 |
+
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/act) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
|
| 18 |
|
| 19 |
Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
|
| 20 |
|
|
|
|
| 27 |
|
| 28 |
| Runtime | Precision | Chipset | SDK Versions | Download |
|
| 29 |
|---|---|---|---|---|
|
| 30 |
+
| ONNX | float | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-onnx-float.zip)
|
| 31 |
+
| QNN_DLC | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-qnn_dlc-float.zip)
|
| 32 |
+
| TFLITE | float | Universal | QAIRT 2.50 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-tflite-float.zip)
|
| 33 |
|
| 34 |
For more device-specific assets and performance metrics, visit **[ACT on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/act)**.
|
| 35 |
|
| 36 |
|
| 37 |
### Option 2: Export with Custom Configurations
|
| 38 |
|
| 39 |
+
Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/act) Python library to compile and export the model with your own:
|
| 40 |
- Custom weights (e.g., fine-tuned checkpoints)
|
| 41 |
- Custom input shapes
|
| 42 |
- Target device and runtime configurations
|
| 43 |
|
| 44 |
This option is ideal if you need to customize the model beyond the default configuration provided here.
|
| 45 |
|
| 46 |
+
See our repository for [ACT on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.63.0/src/qai_hub_models/models/act) for usage instructions.
|
| 47 |
|
| 48 |
## Model Details
|
| 49 |
|
|
|
|
| 58 |
## Performance Summary
|
| 59 |
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|
| 60 |
|---|---|---|---|---|---|---
|
| 61 |
+
| ACT | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 5.917 ms | 1 - 359 MB | NPU
|
| 62 |
+
| ACT | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 6.714 ms | 2 - 373 MB | NPU
|
| 63 |
+
| ACT | ONNX | float | Snapdragon® X2 Elite | 6.62 ms | 4 - 4 MB | NPU
|
| 64 |
+
| ACT | ONNX | float | Snapdragon® X Elite | 12.571 ms | 64 - 64 MB | NPU
|
| 65 |
+
| ACT | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.293 ms | 0 - 465 MB | NPU
|
| 66 |
+
| ACT | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 17.877 ms | 0 - 310 MB | NPU
|
| 67 |
+
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 16.888 ms | 3 - 11 MB | NPU
|
| 68 |
+
| ACT | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 11.821 ms | 0 - 78 MB | NPU
|
| 69 |
+
| ACT | ONNX | float | Qualcomm® QCS8450 | 17.877 ms | 0 - 310 MB | NPU
|
| 70 |
+
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 17.657 ms | 3 - 10 MB | NPU
|
| 71 |
+
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 12.571 ms | 64 - 64 MB | NPU
|
| 72 |
+
| ACT | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 6.714 ms | 2 - 373 MB | NPU
|
| 73 |
+
| ACT | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 4.301 ms | 4 - 381 MB | NPU
|
| 74 |
+
| ACT | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 5.271 ms | 0 - 284 MB | NPU
|
| 75 |
+
| ACT | QNN_DLC | float | Snapdragon® X2 Elite | 5.26 ms | 4 - 4 MB | NPU
|
| 76 |
+
| ACT | QNN_DLC | float | Snapdragon® X Elite | 11.174 ms | 4 - 4 MB | NPU
|
| 77 |
+
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 6.998 ms | 0 - 479 MB | NPU
|
| 78 |
+
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 15.221 ms | 4 - 251 MB | NPU
|
| 79 |
+
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 14.698 ms | 4 - 10 MB | NPU
|
| 80 |
+
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 10.368 ms | 4 - 6 MB | NPU
|
| 81 |
| ACT | QNN_DLC | float | Qualcomm® SA8650P | 14.499 ms | 1 - 271 MB | NPU
|
| 82 |
| ACT | QNN_DLC | float | Qualcomm® SA8255P | 14.499 ms | 1 - 271 MB | NPU
|
| 83 |
+
| ACT | QNN_DLC | float | Qualcomm® QCS8450 | 15.221 ms | 4 - 251 MB | NPU
|
| 84 |
+
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 14.771 ms | 4 - 9 MB | NPU
|
| 85 |
+
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 11.174 ms | 4 - 4 MB | NPU
|
| 86 |
+
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 5.271 ms | 0 - 284 MB | NPU
|
| 87 |
+
| ACT | QNN_DLC | float | Qualcomm® SA8295P | 16.621 ms | 0 - 200 MB | NPU
|
| 88 |
+
| ACT | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 3.838 ms | 0 - 376 MB | NPU
|
| 89 |
+
| ACT | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.757 ms | 0 - 323 MB | NPU
|
| 90 |
+
| ACT | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 6.162 ms | 0 - 469 MB | NPU
|
| 91 |
+
| ACT | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 13.527 ms | 0 - 264 MB | NPU
|
| 92 |
+
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 14.362 ms | 0 - 72 MB | NPU
|
| 93 |
+
| ACT | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.635 ms | 0 - 7 MB | NPU
|
|
|
|
|
|
|
| 94 |
| ACT | TFLITE | float | Qualcomm® SA8650P | 13.799 ms | 0 - 279 MB | NPU
|
| 95 |
| ACT | TFLITE | float | Qualcomm® SA8255P | 13.799 ms | 0 - 279 MB | NPU
|
| 96 |
+
| ACT | TFLITE | float | Qualcomm® QCS8450 | 13.527 ms | 0 - 264 MB | NPU
|
| 97 |
+
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 14.033 ms | 0 - 71 MB | NPU
|
| 98 |
+
| ACT | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.757 ms | 0 - 323 MB | NPU
|
| 99 |
+
| ACT | TFLITE | float | Qualcomm® SA8295P | 15.651 ms | 0 - 198 MB | NPU
|
|
|
|
|
|
|
|
|
|
| 100 |
|
| 101 |
## License
|
| 102 |
* The license for the original implementation of ACT can be found
|
release_assets.json
CHANGED
|
@@ -1,26 +1,26 @@
|
|
| 1 |
{
|
| 2 |
-
"version": "0.
|
| 3 |
"precisions": {
|
| 4 |
"float": {
|
| 5 |
"universal_assets": {
|
| 6 |
"onnx": {
|
| 7 |
"tool_versions": {
|
| 8 |
-
"qairt": "2.
|
| 9 |
"onnx_runtime": "1.27.1"
|
| 10 |
},
|
| 11 |
-
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.
|
| 12 |
},
|
| 13 |
"qnn_dlc": {
|
| 14 |
"tool_versions": {
|
| 15 |
-
"qairt": "2.
|
| 16 |
},
|
| 17 |
-
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.
|
| 18 |
},
|
| 19 |
"tflite": {
|
| 20 |
"tool_versions": {
|
| 21 |
-
"qairt": "2.
|
| 22 |
},
|
| 23 |
-
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.
|
| 24 |
}
|
| 25 |
}
|
| 26 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"version": "0.63.0",
|
| 3 |
"precisions": {
|
| 4 |
"float": {
|
| 5 |
"universal_assets": {
|
| 6 |
"onnx": {
|
| 7 |
"tool_versions": {
|
| 8 |
+
"qairt": "2.50.0.260828221209",
|
| 9 |
"onnx_runtime": "1.27.1"
|
| 10 |
},
|
| 11 |
+
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-onnx-float.zip"
|
| 12 |
},
|
| 13 |
"qnn_dlc": {
|
| 14 |
"tool_versions": {
|
| 15 |
+
"qairt": "2.50.0.260828221209"
|
| 16 |
},
|
| 17 |
+
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-qnn_dlc-float.zip"
|
| 18 |
},
|
| 19 |
"tflite": {
|
| 20 |
"tool_versions": {
|
| 21 |
+
"qairt": "2.50.0.260828221209"
|
| 22 |
},
|
| 23 |
+
"download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/act/releases/v0.63.0/act-tflite-float.zip"
|
| 24 |
}
|
| 25 |
}
|
| 26 |
}
|