Beit: Optimized for Qualcomm Devices

Beit is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.

This is based on the implementation of Beit found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime	Precision	Chipset	SDK Versions	Download
ONNX	float	Universal	QAIRT 2.50, ONNX Runtime 1.30.0	Download
ONNX	w8a16	Universal	QAIRT 2.50, ONNX Runtime 1.30.0	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
QNN_DLC	w8a16	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	Download

For more device-specific assets and performance metrics, visit Beit on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

• Custom weights (e.g., fine-tuned checkpoints)
• Custom input shapes
• Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for Beit on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Input resolution: 224x224
• Model checkpoint: Imagenet
• Model size (float): 351 MB
• Number of parameters: 92.0M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Beit	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.498 ms	1 - 307 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	8.485 ms	0 - 299 MB	NPU
Beit	ONNX	float	Snapdragon® X2 Elite	7.379 ms	2 - 2 MB	NPU
Beit	ONNX	float	Snapdragon® X Elite	15.29 ms	184 - 184 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 3 Mobile	10.605 ms	1 - 445 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 1 Mobile	14.066 ms	1 - 415 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	18.141 ms	0 - 5 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.757 ms	0 - 194 MB	NPU
Beit	ONNX	float	Qualcomm® QCS8450	14.066 ms	1 - 415 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	17.887 ms	0 - 4 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	15.29 ms	184 - 184 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ Q-8750	8.485 ms	0 - 299 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	2.293 ms	0 - 212 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.919 ms	0 - 309 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X2 Elite	2.559 ms	1 - 1 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X Elite	6.669 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	4.566 ms	0 - 417 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	8.147 ms	0 - 418 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	26.724 ms	0 - 4 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.218 ms	0 - 5 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.506 ms	0 - 98 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCS8450	8.147 ms	0 - 418 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	6.3 ms	0 - 4 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.669 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	62.269 ms	0 - 412 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	7.743 ms	0 - 384 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.919 ms	0 - 309 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	7.743 ms	0 - 384 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.421 ms	0 - 299 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	8.457 ms	1 - 294 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X2 Elite	7.719 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X Elite	15.462 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	10.574 ms	0 - 391 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	13.839 ms	0 - 363 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	18.251 ms	1 - 4 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.683 ms	1 - 3 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	17.517 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8650P	17.517 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8255P	17.517 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8450	13.839 ms	0 - 363 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	17.596 ms	3 - 5 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	15.462 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	8.457 ms	1 - 294 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA7255P	48.2 ms	1 - 292 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8295P	14.963 ms	1 - 271 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	2.597 ms	0 - 203 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.427 ms	0 - 302 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.184 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X Elite	7.786 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.839 ms	0 - 407 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.186 ms	0 - 4 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	7.236 ms	0 - 3 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8775P	7.475 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8650P	7.475 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8255P	7.475 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	7.302 ms	2 - 5 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	7.786 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	64.008 ms	0 - 400 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	8.415 ms	0 - 377 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	3.427 ms	0 - 302 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	8.415 ms	0 - 377 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.43 ms	0 - 302 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	8.441 ms	0 - 297 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	10.574 ms	0 - 393 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	13.919 ms	0 - 363 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	18.082 ms	0 - 188 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.343 ms	0 - 3 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	17.526 ms	0 - 294 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8650P	17.526 ms	0 - 294 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8255P	17.526 ms	0 - 294 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8450	13.919 ms	0 - 363 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	17.78 ms	0 - 187 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	8.441 ms	0 - 297 MB	NPU
Beit	TFLITE	float	Qualcomm® SA7255P	48.248 ms	0 - 292 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8295P	15.001 ms	0 - 271 MB	NPU

License

• The license for the original implementation of Beit can be found here.

References

• BEIT: BERT Pre-Training of Image Transformers
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.