Distil-Bert-Base-Uncased-Hf: Optimized for Qualcomm Devices

DistilBERT 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 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	w8a16	Universal	QAIRT 2.45, ONNX Runtime 1.27.1	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 Distil-Bert-Base-Uncased-Hf 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 Distil-Bert-Base-Uncased-Hf on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

• Model checkpoint: distil_bert_base_uncased_hf
• Input resolution: 1x384
• Number of parameters: 11.3M
• Model size (float): 43.3 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® X2 Elite	2.738 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® X Elite	6.153 ms	112 - 112 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	4.189 ms	0 - 367 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	11.671 ms	0 - 369 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	44.574 ms	12 - 15 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	5.906 ms	0 - 114 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® QCS8450	11.671 ms	0 - 369 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	7.072 ms	0 - 3 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.153 ms	112 - 112 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	3.313 ms	0 - 315 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® 8 Elite Mobile	3.313 ms	0 - 315 MB	NPU
Distil-Bert-Base-Uncased-Hf	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.488 ms	0 - 318 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® X2 Elite	7.04 ms	0 - 0 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® X Elite	13.854 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	11.446 ms	0 - 488 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	21.875 ms	0 - 317 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	48.949 ms	0 - 360 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.762 ms	0 - 2 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® SA8775P	18.076 ms	0 - 361 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® SA8650P	18.076 ms	0 - 361 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® SA8255P	18.076 ms	0 - 361 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® QCS8450	21.875 ms	0 - 317 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	17.812 ms	0 - 2 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	13.854 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	7.912 ms	0 - 358 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® SA7255P	48.949 ms	0 - 360 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Qualcomm® SA8295P	23.786 ms	0 - 215 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® 8 Elite Mobile	7.912 ms	0 - 358 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	6.41 ms	0 - 362 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Snapdragon® X2 Elite	3.224 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Snapdragon® X Elite	6.777 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	4.416 ms	0 - 366 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8275	15.601 ms	0 - 308 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	6.303 ms	0 - 172 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® SA8775P	6.502 ms	0 - 309 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® SA8650P	6.502 ms	0 - 309 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® SA8255P	6.502 ms	0 - 309 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	6.283 ms	2 - 4 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	6.777 ms	1 - 1 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	3.53 ms	0 - 313 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Qualcomm® SA7255P	15.601 ms	0 - 308 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	3.53 ms	0 - 313 MB	NPU
Distil-Bert-Base-Uncased-Hf	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2.647 ms	0 - 320 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	11.519 ms	0 - 491 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	21.576 ms	0 - 317 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	49.337 ms	0 - 367 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	14.371 ms	0 - 2 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® SA8775P	18.388 ms	0 - 366 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® SA8650P	18.388 ms	0 - 366 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® SA8255P	18.388 ms	0 - 366 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® QCS8450	21.576 ms	0 - 317 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	18.014 ms	0 - 176 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	8.001 ms	0 - 364 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® SA7255P	49.337 ms	0 - 367 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Qualcomm® SA8295P	23.803 ms	0 - 216 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Snapdragon® 8 Elite Mobile	8.001 ms	0 - 364 MB	NPU
Distil-Bert-Base-Uncased-Hf	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	6.661 ms	0 - 370 MB	NPU

License

• The license for the original implementation of Distil-Bert-Base-Uncased-Hf can be found here.

References

• DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter

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.