EfficientFormer: Optimized for Qualcomm Devices

EfficientFormer is a vision transformer model that can classify images from the Imagenet dataset.

This is based on the implementation of EfficientFormer 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 EfficientFormer 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 EfficientFormer on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Model checkpoint: efficientformer_l1_300d
• Input resolution: 224x224
• Number of parameters: 12.3M
• Model size (float): 46.9 MB
• Model size (w8a16): 12.2 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
EfficientFormer	ONNX	float	Snapdragon® X Elite	1.528 ms	24 - 24 MB	NPU
EfficientFormer	ONNX	float	Snapdragon® 8 Gen 3 Mobile	0.95 ms	0 - 87 MB	NPU
EfficientFormer	ONNX	float	Qualcomm® QCS8550 (Proxy)	1.32 ms	1 - 5 MB	NPU
EfficientFormer	ONNX	float	Qualcomm® QCS9075	1.883 ms	1 - 3 MB	NPU
EfficientFormer	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	0.704 ms	0 - 46 MB	NPU
EfficientFormer	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	0.609 ms	1 - 53 MB	NPU
EfficientFormer	ONNX	float	Snapdragon® X2 Elite	0.646 ms	25 - 25 MB	NPU
EfficientFormer	ONNX	w8a16	Snapdragon® X Elite	1.661 ms	12 - 12 MB	NPU
EfficientFormer	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	0.962 ms	0 - 93 MB	NPU
EfficientFormer	ONNX	w8a16	Qualcomm® QCS6490	143.031 ms	19 - 25 MB	CPU
EfficientFormer	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	1.401 ms	0 - 17 MB	NPU
EfficientFormer	ONNX	w8a16	Qualcomm® QCS9075	1.631 ms	0 - 3 MB	NPU
EfficientFormer	ONNX	w8a16	Qualcomm® QCM6690	66.257 ms	12 - 20 MB	CPU
EfficientFormer	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	0.647 ms	0 - 63 MB	NPU
EfficientFormer	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	62.871 ms	22 - 31 MB	CPU
EfficientFormer	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	0.531 ms	0 - 66 MB	NPU
EfficientFormer	ONNX	w8a16	Snapdragon® X2 Elite	0.562 ms	12 - 12 MB	NPU
EfficientFormer	QNN_DLC	float	Snapdragon® X Elite	1.744 ms	1 - 1 MB	NPU
EfficientFormer	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.063 ms	0 - 80 MB	NPU
EfficientFormer	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	4.912 ms	1 - 45 MB	NPU
EfficientFormer	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	1.498 ms	1 - 2 MB	NPU
EfficientFormer	QNN_DLC	float	Qualcomm® QCS9075	1.979 ms	1 - 3 MB	NPU
EfficientFormer	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	5.532 ms	0 - 83 MB	NPU
EfficientFormer	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	0.776 ms	1 - 48 MB	NPU
EfficientFormer	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	0.649 ms	1 - 49 MB	NPU
EfficientFormer	QNN_DLC	float	Snapdragon® X2 Elite	0.919 ms	1 - 1 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® X Elite	1.846 ms	0 - 0 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	1.095 ms	0 - 81 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	3.289 ms	0 - 56 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	1.62 ms	0 - 89 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Qualcomm® QCS9075	1.785 ms	0 - 2 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Qualcomm® QCM6690	7.002 ms	0 - 176 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	0.719 ms	0 - 55 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	1.722 ms	0 - 60 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	0.585 ms	0 - 59 MB	NPU
EfficientFormer	QNN_DLC	w8a16	Snapdragon® X2 Elite	0.831 ms	0 - 0 MB	NPU
EfficientFormer	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1.048 ms	0 - 104 MB	NPU
EfficientFormer	TFLITE	float	Qualcomm® QCS8275 (Proxy)	4.91 ms	0 - 64 MB	NPU
EfficientFormer	TFLITE	float	Qualcomm® QCS8550 (Proxy)	1.505 ms	0 - 4 MB	NPU
EfficientFormer	TFLITE	float	Qualcomm® QCS9075	1.969 ms	0 - 27 MB	NPU
EfficientFormer	TFLITE	float	Qualcomm® QCS8450 (Proxy)	5.534 ms	0 - 101 MB	NPU
EfficientFormer	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	0.777 ms	0 - 70 MB	NPU
EfficientFormer	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	0.651 ms	0 - 68 MB	NPU

License

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

References

• Rethinking Vision Transformers for MobileNet Size and Speed
• 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.