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 2.42, ONNX Runtime 1.24.3	Download
ONNX	w8a16	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Beit	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	7.254 ms	1 - 464 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite Mobile	8.594 ms	1 - 465 MB	NPU
Beit	ONNX	float	Snapdragon® X2 Elite	7.425 ms	184 - 184 MB	NPU
Beit	ONNX	float	Snapdragon® X Elite	15.337 ms	183 - 183 MB	NPU
Beit	ONNX	float	Snapdragon® X Elite	15.337 ms	183 - 183 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 3 Mobile	10.575 ms	0 - 505 MB	NPU
Beit	ONNX	float	Qualcomm® QCS8550 (Proxy)	14.622 ms	0 - 201 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	8.594 ms	1 - 465 MB	NPU
Beit	ONNX	float	Qualcomm® QCS9075	18.108 ms	0 - 4 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	3.588 ms	0 - 282 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite Mobile	5.67 ms	0 - 385 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X2 Elite	3.877 ms	101 - 101 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X Elite	10.398 ms	102 - 102 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X Elite	10.398 ms	102 - 102 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	6.685 ms	0 - 394 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCS6490	1066.557 ms	50 - 69 MB	CPU
Beit	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	9.652 ms	0 - 117 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	5.67 ms	0 - 385 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	581.89 ms	75 - 90 MB	CPU
Beit	ONNX	w8a16	Qualcomm® QCM6690	605.466 ms	77 - 89 MB	CPU
Beit	ONNX	w8a16	Qualcomm® QCS9075	10.837 ms	0 - 3 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	581.89 ms	75 - 90 MB	CPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	7.155 ms	1 - 323 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite Mobile	8.57 ms	0 - 323 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X2 Elite	7.887 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X Elite	15.142 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X Elite	15.142 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	10.478 ms	1 - 411 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	14.321 ms	1 - 3 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	17.182 ms	1 - 319 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	17.182 ms	1 - 319 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	17.182 ms	1 - 319 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA7255P	47.995 ms	1 - 320 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	17.058 ms	0 - 420 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8295P	14.604 ms	1 - 327 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	8.57 ms	0 - 323 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS9075	18.289 ms	1 - 3 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.737 ms	0 - 277 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	4.068 ms	0 - 369 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.412 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X Elite	7.71 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X Elite	7.71 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.77 ms	0 - 442 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	7.163 ms	0 - 2 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8775P	7.355 ms	0 - 382 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8775P	7.355 ms	0 - 382 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8775P	7.355 ms	0 - 382 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA7255P	15.918 ms	0 - 381 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.068 ms	0 - 369 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	8.666 ms	0 - 409 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® QCM6690	65.619 ms	0 - 439 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® QCS9075	7.924 ms	0 - 2 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	8.666 ms	0 - 409 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	7.16 ms	0 - 308 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite Mobile	8.596 ms	0 - 311 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	10.479 ms	0 - 397 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8550 (Proxy)	14.092 ms	0 - 2 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	17.166 ms	0 - 300 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	17.166 ms	0 - 300 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	17.166 ms	0 - 300 MB	NPU
Beit	TFLITE	float	Qualcomm® SA7255P	48.036 ms	0 - 301 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8450 (Proxy)	17.579 ms	0 - 370 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8295P	15.34 ms	0 - 294 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	8.596 ms	0 - 311 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS9075	18.616 ms	0 - 186 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.