Modelpublic

qualcomm/Llama-v3-8B-Instruct

sourceHugging Facellama3updated 11d agoView on Hugging Face

0likes

Model Card

Llama-v3-8B-Instruct: Optimized for Qualcomm Devices

Llama 3 is a family of LLMs. The model is quantized to w4a16 (4-bit weights and 16-bit activations) and part of the model is quantized to w8a16 (8-bit weights and 16-bit activations) making it suitable for on-device deployment. For Prompt and output length specified below, the time to first token is Llama-PromptProcessor-Quantized's latency and average time per addition token is Llama-TokenGenerator-Quantized's latency.

This is based on the implementation of Llama-v3-8B-Instruct 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 Llama-v3-8B-Instruct on-device

Follow the GenieX quickstart to install GenieX and deploy the model on a target device.

You'll need to export the model artifact using the steps below, then follow Run a Local Model with GenieX.

See the LLM-on-Genie tutorial to run with the Genie runtime. Note: Genie support will be deprecated soon.

Getting Started

Due to licensing restrictions, we cannot distribute pre-exported model assets for this model. 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

See our repository for Llama-v3-8B-Instruct on GitHub for usage instructions.

Model Details

Model Type: Modelusecase.text_generation

Model Stats:

—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. The lower bound is for a short prompt (up to 128 tokens, i.e., one iteration of the prompt processor) and the upper bound is for a prompt using the full context length (4096 tokens).

Performance Summary

Model	Runtime	Precision	Chipset	Context Length	Response Rate (tokens per second)	Time To First Token (range, seconds)
Llama-v3-8B-Instruct	GENIE	w4a16	Snapdragon® 8 Elite Gen 5 Mobile	4096	16.36505870819092	0.0993634 - 3.1796288
Llama-v3-8B-Instruct	GENIE	w4a16	Snapdragon® 8 Elite Mobile	4096	15.001426887512206	0.1370342 - 4.3850944
Llama-v3-8B-Instruct	GENIE	w4a16	Snapdragon® X2 Elite	4096	19.47	0.147975 - 4.7352
Llama-v3-8B-Instruct	GENIE	w4a16	Snapdragon® X Elite	4096	4.642633438110352	0.20893 - 6.68576
Llama-v3-8B-Instruct	GENIE	w4a16	Qualcomm® Dragonwing™ IQ-9075	4096	10.764844226837159	0.18326900000000002 - 5.8646080000000005
Llama-v3-8B-Instruct	GENIE	w4a16	Qualcomm® Dragonwing™ IQ-X7181	4096	4.642633438110352	0.20893 - 6.68576
Llama-v3-8B-Instruct	GENIE	w4a16	Qualcomm® Dragonwing™ Q-8750	4096	15.001426887512206	0.1370342 - 4.3850944
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Snapdragon® 8 Elite Gen 5 Mobile	4096	14.343045	0.1397824193548387 - 4.473037419354839
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Snapdragon® 8 Elite Mobile	4096	14.052707	0.182029 - 5.824928
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Snapdragon® X2 Elite	4096	21.24863	0.11502590322580644 - 3.680828903225806
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Snapdragon® X Elite	4096	11.330596	0.22880099999999998 - 7.321631999999999
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Qualcomm® Dragonwing™ IQ-8275	4096	9.875264	0.220166 - 7.045312
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Qualcomm® Dragonwing™ IQ-9075	4096	9.6774	0.20317929032258064 - 6.5017372903225805
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Qualcomm® Dragonwing™ IQ-X7181	4096	11.330596	0.22880099999999998 - 7.321631999999999
Llama-v3-8B-Instruct	GENIEX_QAIRT	w4a16	Qualcomm® Dragonwing™ Q-8750	4096	14.052707	0.182029 - 5.824928

License

—The license for the original implementation of Llama-v3-8B-Instruct can be found here.

References

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.

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