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 1.24.1	Download
ONNX	w8a16	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a16	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	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® X2 Elite	5.944 ms	185 - 185 MB	NPU
Beit	ONNX	float	Snapdragon® X Elite	13.666 ms	185 - 185 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 3 Mobile	9.284 ms	0 - 532 MB	NPU
Beit	ONNX	float	Qualcomm® QCS8550 (Proxy)	12.909 ms	0 - 195 MB	NPU
Beit	ONNX	float	Qualcomm® QCS9075	17.775 ms	0 - 4 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	6.604 ms	0 - 482 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	6.236 ms	1 - 487 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X2 Elite	4.367 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X Elite	12.501 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	7.962 ms	0 - 495 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCS6490	1069.347 ms	51 - 68 MB	CPU
Beit	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	11.726 ms	0 - 116 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCS9075	14.685 ms	0 - 3 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCM6690	601.942 ms	67 - 78 MB	CPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	6.119 ms	0 - 407 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	582.238 ms	114 - 127 MB	CPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	4.182 ms	0 - 407 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X2 Elite	6.976 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X Elite	13.417 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	8.724 ms	0 - 531 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	44.941 ms	1 - 485 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	12.659 ms	1 - 4 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	16.408 ms	1 - 485 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS9075	17.052 ms	3 - 5 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	23.058 ms	0 - 507 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA7255P	44.941 ms	1 - 485 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8295P	19.086 ms	1 - 468 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	6.867 ms	1 - 478 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	6.57 ms	1 - 481 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	6.664 ms	0 - 348 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8275 (Proxy)	38.588 ms	0 - 298 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8550 (Proxy)	9.321 ms	0 - 3 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	12.146 ms	0 - 306 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS9075	13.563 ms	0 - 187 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8450 (Proxy)	19.275 ms	0 - 431 MB	NPU
Beit	TFLITE	float	Qualcomm® SA7255P	38.588 ms	0 - 298 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8295P	15.994 ms	0 - 405 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	4.764 ms	0 - 300 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.967 ms	0 - 297 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.