Hugging Face's logo

qualcomm
/
Granite-4.0-Micro

Text Generation
PyTorch
llm
generative_ai
android
Model card Files
xet
 Community

Granite-4.0-Micro: Optimized for Qualcomm Devices

Granite 4.0 is a family of open language models from IBM designed for enterprise AI workloads including code generation, summarization, and retrieval-augmented generation.

This is based on the implementation of Granite-4.0-Micro 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 Granite-4.0-Micro on-device

Follow the GenieX quickstart to install GenieX and deploy the model on a target 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
GENIEX_LLAMACPP q4_0 Universal Download

For more device-specific assets and performance metrics, visit Granite-4.0-Micro 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 Granite-4.0-Micro on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

  • Model architecture: Mamba-2 Hybrid architecture combining State Space Model (SSM) layers with Attention layers for efficient long-context processing.
  • Response Rate: Rate of response generation after the first response token.
  • Supported languages: English
  • TTFT: Time To First Token is the time it takes to generate the first response token. This is expressed as a range because it varies based on the length of the prompt.

Performance Summary

Model Runtime Precision Chipset Context Length Response Rate (tokens per second) Time To First Token (range, seconds)
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 25.511201 0.7986566666666666 - 3.1946266666666663
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 22.446851 0.49678866666666666 - 1.9871546666666666
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 18.111545 0.19774833333333333 - 0.7909933333333333
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 23.532299 1.1548540967741936 - 36.955331096774195
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 17.397706 0.8204811290322581 - 26.25539612903226
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 13.732114 0.29203270967741934 - 9.345046709677419
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 43.451812 0.2649713333333333 - 1.0598853333333331
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 21.465781 0.27547299999999997 - 1.1018919999999999
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 15.783036 0.16830266666666666 - 0.6732106666666666
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 40.183774 0.2600884838709678 - 8.322831483870969
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 18.901314 0.2853414838709677 - 9.130927483870966
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 12.796386 0.16912774193548386 - 5.412087741935483
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 19.042178 0.8285933333333334 - 3.3143733333333336
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 24.946346 0.5435570000000001 - 2.1742280000000003
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 7.57187 0.45771866666666666 - 1.8308746666666667
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 28.578077 0.7872631935483871 - 25.19242219354839
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 21.913417 0.5831270322580644 - 18.66006503225806
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 6.746864 0.6111447741935484 - 19.55663277419355
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 8.884934 3.9245596666666667 - 15.698238666666667
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 3.53597 12.026639666666666 - 48.106558666666665
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 5.443207 1.1636513333333331 - 4.654605333333333
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 7.190223 4.759664580645161 - 152.30926658064516
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 3.394532 12.195987483870967 - 390.27159948387094
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 5.587816 1.1214709677419354 - 35.887070967741934
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 19.042178 0.8285933333333334 - 3.3143733333333336
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 24.946346 0.5435570000000001 - 2.1742280000000003
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 7.57187 0.45771866666666666 - 1.8308746666666667
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 27.25256 0.7220270000000001 - 2.8881080000000003
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 8.452642 0.641794 - 2.567176
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 14.889491 0.24504266666666666 - 0.9801706666666666
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 28.578077 0.7872631935483871 - 25.19242219354839
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 21.913417 0.5831270322580644 - 18.66006503225806
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 6.746864 0.6111447741935484 - 19.55663277419355
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 24.078399 0.9540022580645161 - 30.528072258064515
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 8.394046 0.6556692903225807 - 20.981417290322582
Granite-4.0-Micro GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 13.762181 0.2654138709677419 - 8.49324387096774

License

  • The license for the original implementation of Granite-4.0-Micro can be found here.

References

Community

Usage and Limitations

This model may not be used for or in connection with any of the following applications:

  • Accessing essential private and public services and benefits;
  • Administration of justice and democratic processes;
  • Assessing or recognizing the emotional state of a person;
  • Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
  • Education and vocational training;
  • Employment and workers management;
  • Exploitation of the vulnerabilities of persons resulting in harmful behavior;
  • General purpose social scoring;
  • Law enforcement;
  • Management and operation of critical infrastructure;
  • Migration, asylum and border control management;
  • Predictive policing;
  • Real-time remote biometric identification in public spaces;
  • Recommender systems of social media platforms;
  • Scraping of facial images (from the internet or otherwise); and/or
  • Subliminal manipulation
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