Hugging Face's logo

qualcomm
/
OpusMT-Ko-En

Text Generation
PyTorch
foundation
android
Model card Files
xet
 Community

OpusMT-Ko-En: Optimized for Qualcomm Devices

OpusMT Korean to English translation model is a state-of-the-art neural machine translation system designed for translating Korean text into English. This model is based on the Marian transformer architecture and has been optimized for edge inference by splitting into encoder and decoder components with modified attention mechanisms. It exhibits robust performance for real-world translation tasks, making it highly reliable for practical applications. The model supports input sequences up to 256 tokens and can generate Korean translations with high accuracy.

This is based on the implementation of OpusMT-Ko-En 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.

Deploying OpusMT-Ko-En on-device

This model is compatible with the Qualcomm Voice AI SDK. Download the SDK from the Qualcomm Package Manager to deploy this model on-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
VOICE_AI float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile QAIRT 2.50 Download
VOICE_AI float Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.50 Download
VOICE_AI float Snapdragon® X2 Elite QAIRT 2.50 Download
VOICE_AI float Snapdragon® X Elite QAIRT 2.50 Download
VOICE_AI float Snapdragon® 8 Gen 1 Mobile QAIRT 2.50 Download
VOICE_AI float Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.50 Download
VOICE_AI float Qualcomm® Dragonwing™ QCS8550 (Proxy) QAIRT 2.50 Download
VOICE_AI float Qualcomm® SA8775P QAIRT 2.50 Download
VOICE_AI float Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.50 Download
VOICE_AI float Qualcomm® SA7255P QAIRT 2.50 Download
VOICE_AI float Qualcomm® SA8295P QAIRT 2.50 Download
VOICE_AI float Snapdragon® 8 Elite Extreme Gen 6 Mobile QAIRT 2.50 Download
VOICE_AI float Snapdragon® 8 Elite Gen 6 Mobile QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® X2 Elite QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® X Elite QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® 8 Gen 1 Mobile QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® SA8775P QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® SA7255P QAIRT 2.50 Download
VOICE_AI w8a16 Qualcomm® SA8295P QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® 8 Elite Extreme Gen 6 Mobile QAIRT 2.50 Download
VOICE_AI w8a16 Snapdragon® 8 Elite Gen 6 Mobile QAIRT 2.50 Download

For more device-specific assets and performance metrics, visit OpusMT-Ko-En 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 OpusMT-Ko-En on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

  • Attention heads: 8
  • Hidden dimension: 512
  • Input resolution: 256 tokens (Korean text)
  • Max input sequence length: 256 tokens
  • Max output sequence length: 256 tokens
  • Model checkpoint: Helsinki-NLP/opus-mt-ko-en
  • Model size (decoder) (float): 340 MB
  • Model size (decoder) (w8a16): 90.8 MB
  • Model size (encoder) (float): 212 MB
  • Model size (encoder) (w8a16): 55.2 MB
  • Number of decoder layers: 6
  • Number of encoder layers: 6
  • Number of parameters (decoder): 55.7M
  • Number of parameters (encoder): 55.5M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
decoder VOICE_AI float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1.362 ms 0 - 9 MB NPU
decoder VOICE_AI float Snapdragon® 8 Elite For Galaxy Mobile 1.475 ms 0 - 12 MB NPU
decoder VOICE_AI float Snapdragon® X2 Elite 1.327 ms 12 - 12 MB NPU
decoder VOICE_AI float Snapdragon® X Elite 2.162 ms 12 - 12 MB NPU
decoder VOICE_AI float Snapdragon® 8 Gen 1 Mobile 2.884 ms 12 - 21 MB NPU
decoder VOICE_AI float Qualcomm® Dragonwing™ IQ-8275 3.02 ms 12 - 27 MB NPU
decoder VOICE_AI float Qualcomm® Dragonwing™ QCS8550 (Proxy) 2.338 ms 3 - 4 MB NPU
decoder VOICE_AI float Qualcomm® SA8775P 3.006 ms 12 - 21 MB NPU
decoder VOICE_AI float Qualcomm® SA8650P 3.006 ms 12 - 21 MB NPU
decoder VOICE_AI float Qualcomm® SA8255P 3.006 ms 12 - 21 MB NPU
decoder VOICE_AI float Qualcomm® QCS8450 2.884 ms 12 - 21 MB NPU
decoder VOICE_AI float Qualcomm® Dragonwing™ IQ-9075 2.851 ms 12 - 27 MB NPU
decoder VOICE_AI float Qualcomm® Dragonwing™ IQ-X7181 2.162 ms 12 - 12 MB NPU
decoder VOICE_AI float Qualcomm® Dragonwing™ Q-8750 1.475 ms 0 - 12 MB NPU
decoder VOICE_AI float Qualcomm® SA7255P 3.977 ms 12 - 20 MB NPU
decoder VOICE_AI float Qualcomm® SA8295P 3.14 ms 10 - 15 MB NPU
decoder VOICE_AI float Snapdragon® 8 Elite Extreme Gen 6 Mobile 1.335 ms 1 - 11 MB NPU
decoder VOICE_AI float Snapdragon® 8 Elite Gen 6 Mobile 1.347 ms 1 - 11 MB NPU
decoder VOICE_AI w8a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 0.962 ms 0 - 8 MB NPU
decoder VOICE_AI w8a16 Snapdragon® 8 Elite For Galaxy Mobile 1.007 ms 0 - 12 MB NPU
decoder VOICE_AI w8a16 Snapdragon® X2 Elite 0.902 ms 3 - 3 MB NPU
decoder VOICE_AI w8a16 Snapdragon® X Elite 1.465 ms 3 - 3 MB NPU
decoder VOICE_AI w8a16 Snapdragon® 8 Gen 1 Mobile 1.713 ms 1 - 10 MB NPU
decoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-8275 1.9 ms 0 - 5 MB NPU
decoder VOICE_AI w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 1.532 ms 0 - 1 MB NPU
decoder VOICE_AI w8a16 Qualcomm® SA8775P 1.991 ms 0 - 9 MB NPU
decoder VOICE_AI w8a16 Qualcomm® SA8650P 1.991 ms 0 - 9 MB NPU
decoder VOICE_AI w8a16 Qualcomm® SA8255P 1.991 ms 0 - 9 MB NPU
decoder VOICE_AI w8a16 Qualcomm® QCS8450 1.713 ms 1 - 10 MB NPU
decoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-9075 1.821 ms 4 - 7 MB NPU
decoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-X7181 1.465 ms 3 - 3 MB NPU
decoder VOICE_AI w8a16 Qualcomm® Dragonwing™ Q-8750 1.007 ms 0 - 12 MB NPU
decoder VOICE_AI w8a16 Qualcomm® SA7255P 2.841 ms 0 - 8 MB NPU
decoder VOICE_AI w8a16 Qualcomm® SA8295P 2.167 ms 3 - 8 MB NPU
decoder VOICE_AI w8a16 Snapdragon® 8 Elite Extreme Gen 6 Mobile 0.93 ms 0 - 10 MB NPU
decoder VOICE_AI w8a16 Snapdragon® 8 Elite Gen 6 Mobile 0.916 ms 0 - 10 MB NPU
encoder VOICE_AI float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2.602 ms 0 - 9 MB NPU
encoder VOICE_AI float Snapdragon® 8 Elite For Galaxy Mobile 2.953 ms 0 - 12 MB NPU
encoder VOICE_AI float Snapdragon® X2 Elite 2.653 ms 0 - 0 MB NPU
encoder VOICE_AI float Snapdragon® X Elite 4.937 ms 0 - 0 MB NPU
encoder VOICE_AI float Snapdragon® 8 Gen 1 Mobile 6.241 ms 0 - 10 MB NPU
encoder VOICE_AI float Qualcomm® Dragonwing™ IQ-8275 5.082 ms 0 - 9 MB NPU
encoder VOICE_AI float Qualcomm® Dragonwing™ QCS8550 (Proxy) 4.947 ms 0 - 1 MB NPU
encoder VOICE_AI float Qualcomm® SA8775P 5.246 ms 0 - 9 MB NPU
encoder VOICE_AI float Qualcomm® SA8650P 5.246 ms 0 - 9 MB NPU
encoder VOICE_AI float Qualcomm® SA8255P 5.246 ms 0 - 9 MB NPU
encoder VOICE_AI float Qualcomm® QCS8450 6.241 ms 0 - 10 MB NPU
encoder VOICE_AI float Qualcomm® Dragonwing™ IQ-9075 5.063 ms 2 - 10 MB NPU
encoder VOICE_AI float Qualcomm® Dragonwing™ IQ-X7181 4.937 ms 0 - 0 MB NPU
encoder VOICE_AI float Qualcomm® Dragonwing™ Q-8750 2.953 ms 0 - 12 MB NPU
encoder VOICE_AI float Qualcomm® SA7255P 8.331 ms 0 - 9 MB NPU
encoder VOICE_AI float Qualcomm® SA8295P 6.38 ms 0 - 5 MB NPU
encoder VOICE_AI float Snapdragon® 8 Elite Extreme Gen 6 Mobile 1.149 ms 0 - 10 MB NPU
encoder VOICE_AI float Snapdragon® 8 Elite Gen 6 Mobile 2.463 ms 0 - 9 MB NPU
encoder VOICE_AI w8a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2.385 ms 0 - 7 MB NPU
encoder VOICE_AI w8a16 Snapdragon® 8 Elite For Galaxy Mobile 2.796 ms 0 - 8 MB NPU
encoder VOICE_AI w8a16 Snapdragon® X2 Elite 2.301 ms 0 - 0 MB NPU
encoder VOICE_AI w8a16 Snapdragon® X Elite 4.667 ms 0 - 0 MB NPU
encoder VOICE_AI w8a16 Snapdragon® 8 Gen 1 Mobile 5.456 ms 0 - 10 MB NPU
encoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-8275 4.633 ms 0 - 4 MB NPU
encoder VOICE_AI w8a16 Qualcomm® Dragonwing™ QCS8550 (Proxy) 4.636 ms 0 - 3 MB NPU
encoder VOICE_AI w8a16 Qualcomm® SA8775P 4.778 ms 0 - 9 MB NPU
encoder VOICE_AI w8a16 Qualcomm® SA8650P 4.778 ms 0 - 9 MB NPU
encoder VOICE_AI w8a16 Qualcomm® SA8255P 4.778 ms 0 - 9 MB NPU
encoder VOICE_AI w8a16 Qualcomm® QCS8450 5.456 ms 0 - 10 MB NPU
encoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-9075 4.551 ms 0 - 4 MB NPU
encoder VOICE_AI w8a16 Qualcomm® Dragonwing™ IQ-X7181 4.667 ms 0 - 0 MB NPU
encoder VOICE_AI w8a16 Qualcomm® Dragonwing™ Q-8750 2.796 ms 0 - 8 MB NPU
encoder VOICE_AI w8a16 Qualcomm® SA7255P 7.816 ms 0 - 10 MB NPU
encoder VOICE_AI w8a16 Qualcomm® SA8295P 6.007 ms 0 - 5 MB NPU
encoder VOICE_AI w8a16 Snapdragon® 8 Elite Extreme Gen 6 Mobile 0.923 ms 0 - 10 MB NPU
encoder VOICE_AI w8a16 Snapdragon® 8 Elite Gen 6 Mobile 2.191 ms 0 - 10 MB NPU

License

  • The license for the original implementation of OpusMT-Ko-En can be found here.

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Text Generation
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support