README.md

---
library_name: pytorch
license: other
pipeline_tag: text-generation
tags:
- llm
- generative_ai
- quantized
- android

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/jais_6p7b_chat_quantized/web-assets/model_demo.png)

# JAIS-6p7b-Chat: Optimized for Mobile Deployment
## State-of-the-art large language model useful on a variety of language understanding and generation tasks


JAIS 6.7B is a bilingual large language model (LLM) for both Arabic and English developed by Inception, a G42 company in partnership with MBZUAI and Cerebras. This is a 6.7 billion parameter LLM, trained on a dataset containing 141 billion Arabic tokens and 339 billion English/code tokens. The model is based on transformer-based decoder-only (GPT-3) architecture and uses SwiGLU non-linearity. It implements ALiBi position embeddings, enabling the model to extrapolate to long sequence lengths, providing improved context handling and model precision. The JAIS family of models is a comprehensive series of bilingual English-Arabic LLMs. These models are optimized to excel in Arabic while having strong English capabilities.

This model is an implementation of JAIS-6p7b-Chat found [here](https://huggingface.co/inceptionai/jais-family-6p7b).


Please contact us to purchase this model. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/jais_6p7b_chat_quantized).

### Model Details

- **Model Type:** Text generation
- **Model Stats:**
  - Input sequence length for Prompt Processor: 128
  - Max context length: 2048
  - Number of parameters: 6.7B
  - Precision: w4a16 + w8a16 (a few layers)
  - Use: Initiate conversation with prompt-processor and then token generator for subsequent iterations.
  - Supported languages: Arabic (MSA) and English.
  - Minimum QNN SDK version required: 2.27.7
  - 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 (2048 tokens).
  - Response Rate: Rate of response generation after the first response token.

| Model | Device | Chipset | Target Runtime | Response Rate (tokens per second) | Time To First Token (range, seconds)
|---|---|---|---|---|---|
| Jais-6p7b-Chat | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 13.33 | 0.238231 - 3.811696 | -- | -- |

## Deploying JAIS-6p7b-Chat on-device

Please follow the [LLM on-device deployment](https://github.com/quic/ai-hub-apps/tree/main/tutorials/llm_on_genie) tutorial.




## References
* [Jais and Jais-chat: Arabic-Centric Foundation and Instruction-Tuned Open Generative Large Language Models](https://arxiv.org/abs/2308.16149)
* [Source Model Implementation](https://huggingface.co/inceptionai/jais-family-6p7b)



## Community
* Join [our AI Hub Slack community](https://qualcomm-ai-hub.slack.com/join/shared_invite/zt-2d5zsmas3-Sj0Q9TzslueCjS31eXG2UA#/shared-invite/email) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:[email protected]).

## Usage and Limitations

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