Mobile-Bert-Uncased-Google: Optimized for Qualcomm Devices

MOBILEBERT is a lightweight BERT model designed for efficient self-supervised learning of language representations. It can be used for masked language modeling and as a backbone for various NLP tasks.

This is based on the implementation of Mobile-Bert-Uncased-Google 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 2.45, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit Mobile-Bert-Uncased-Google 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 Mobile-Bert-Uncased-Google on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

• Model checkpoint: mobile_bert_uncased_google
• Input resolution: 1x384
• Number of parameters: 25.3M
• Model size (float): 130 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® X2 Elite	15.447 ms	1 - 1 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® X Elite	25.382 ms	80 - 80 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® 8 Gen 3 Mobile	19.56 ms	0 - 365 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® 8 Gen 1 Mobile	30.24 ms	0 - 407 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	26.131 ms	0 - 3 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Qualcomm® QCS8450	30.24 ms	0 - 407 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	39.599 ms	0 - 3 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	25.382 ms	80 - 80 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Qualcomm® Dragonwing™ Q-8750	15.749 ms	0 - 185 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® 8 Elite Mobile	15.749 ms	0 - 185 MB	NPU
Mobile-Bert-Uncased-Google	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	13.939 ms	0 - 186 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® X2 Elite	14.694 ms	1 - 1 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® X Elite	26.043 ms	1 - 1 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	19.785 ms	0 - 310 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	29.81 ms	0 - 380 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	56.737 ms	0 - 195 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	26.272 ms	0 - 2 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® SA8775P	28.637 ms	0 - 195 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® SA8650P	28.637 ms	0 - 195 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® SA8255P	28.637 ms	0 - 195 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® QCS8450	29.81 ms	0 - 380 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	28.267 ms	0 - 2 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	26.043 ms	1 - 1 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	15.574 ms	0 - 213 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® SA7255P	56.737 ms	0 - 195 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Qualcomm® SA8295P	32.966 ms	0 - 260 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® 8 Elite Mobile	15.574 ms	0 - 213 MB	NPU
Mobile-Bert-Uncased-Google	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	14.052 ms	0 - 216 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	19.36 ms	0 - 311 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	28.875 ms	0 - 364 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	25.431 ms	0 - 3 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® SA8775P	27.795 ms	0 - 200 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® SA8650P	27.795 ms	0 - 200 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® SA8255P	27.795 ms	0 - 200 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® QCS8450	28.875 ms	0 - 364 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	27.473 ms	0 - 82 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	14.725 ms	0 - 221 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Qualcomm® SA8295P	32.391 ms	0 - 252 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Snapdragon® 8 Elite Mobile	14.725 ms	0 - 221 MB	NPU
Mobile-Bert-Uncased-Google	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	13.657 ms	0 - 221 MB	NPU

License

• The license for the original implementation of Mobile-Bert-Uncased-Google can be found here.

References

• MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
• 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.