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 1.27.1	Download
ONNX	w8a16	Universal	QAIRT 2.50, ONNX Runtime 1.27.1	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	7.502 ms	1 - 306 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	8.505 ms	1 - 300 MB	NPU
Beit	ONNX	float	Snapdragon® X2 Elite	7.407 ms	2 - 2 MB	NPU
Beit	ONNX	float	Snapdragon® X Elite	15.334 ms	184 - 184 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 3 Mobile	10.57 ms	0 - 445 MB	NPU
Beit	ONNX	float	Snapdragon® 8 Gen 1 Mobile	14.004 ms	1 - 416 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	18.371 ms	0 - 5 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.777 ms	0 - 195 MB	NPU
Beit	ONNX	float	Qualcomm® QCS8450	14.004 ms	1 - 416 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	17.732 ms	1 - 4 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	15.334 ms	184 - 184 MB	NPU
Beit	ONNX	float	Qualcomm® Dragonwing™ Q-8750	8.505 ms	1 - 300 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	2.299 ms	0 - 213 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	4.003 ms	0 - 311 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X2 Elite	2.56 ms	1 - 1 MB	NPU
Beit	ONNX	w8a16	Snapdragon® X Elite	6.702 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	4.588 ms	0 - 416 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	8.279 ms	0 - 419 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	27.042 ms	0 - 4 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.215 ms	0 - 5 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.529 ms	0 - 99 MB	NPU
Beit	ONNX	w8a16	Qualcomm® QCS8450	8.279 ms	0 - 419 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	6.289 ms	0 - 4 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.702 ms	96 - 96 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-6690	61.891 ms	0 - 412 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	7.756 ms	0 - 387 MB	NPU
Beit	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	4.003 ms	0 - 311 MB	NPU
Beit	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	7.756 ms	0 - 387 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.439 ms	3 - 300 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	8.516 ms	1 - 295 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X2 Elite	7.749 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® X Elite	15.538 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	10.595 ms	0 - 391 MB	NPU
Beit	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	13.955 ms	0 - 361 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	18.038 ms	1 - 4 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.694 ms	1 - 4 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8775P	17.495 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8650P	17.495 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8255P	17.495 ms	1 - 293 MB	NPU
Beit	QNN_DLC	float	Qualcomm® QCS8450	13.955 ms	0 - 361 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	17.637 ms	1 - 3 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	15.538 ms	1 - 1 MB	NPU
Beit	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	8.516 ms	1 - 295 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA7255P	48.206 ms	1 - 292 MB	NPU
Beit	QNN_DLC	float	Qualcomm® SA8295P	14.959 ms	1 - 271 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	2.593 ms	0 - 201 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	3.392 ms	0 - 300 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.277 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® X Elite	7.757 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.833 ms	0 - 410 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	6.193 ms	0 - 3 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	7.243 ms	0 - 2 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8775P	7.47 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8650P	7.47 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA8255P	7.47 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	7.292 ms	0 - 3 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	7.757 ms	0 - 0 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	71.563 ms	0 - 400 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	8.363 ms	0 - 377 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	3.392 ms	0 - 300 MB	NPU
Beit	QNN_DLC	w8a16	Qualcomm® SA7255P	14.958 ms	0 - 349 MB	NPU
Beit	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	8.363 ms	0 - 377 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	7.432 ms	0 - 304 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	8.46 ms	0 - 296 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	10.617 ms	0 - 395 MB	NPU
Beit	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	13.931 ms	0 - 364 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	18.264 ms	0 - 188 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.333 ms	0 - 2 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8775P	17.514 ms	0 - 293 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8650P	17.514 ms	0 - 293 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8255P	17.514 ms	0 - 293 MB	NPU
Beit	TFLITE	float	Qualcomm® QCS8450	13.931 ms	0 - 364 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	17.724 ms	0 - 187 MB	NPU
Beit	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	8.46 ms	0 - 296 MB	NPU
Beit	TFLITE	float	Qualcomm® SA7255P	48.246 ms	0 - 293 MB	NPU
Beit	TFLITE	float	Qualcomm® SA8295P	14.985 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.