README.md

---
library_name: pytorch
license: apache-2.0
pipeline_tag: text-generation
tags:
- llm
- generative_ai
- quantized
- android

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/qwen2_7b_instruct_quantized/web-assets/model_demo.png)

# Qwen2-7B-Instruct: Optimized for Mobile Deployment
## State-of-the-art large language model useful on a variety of language understanding and generation tasks


The Qwen2-7B-Instruct is a state-of-the-art multilingual language model with 7.07 billion parameters, excelling in language understanding, generation, coding, and mathematics. AI Hub provides with four QNN context binaries (shared weights) that can be deployed on Snapdragon 8 Elite with Genie SDK.

This model is an implementation of Qwen2-7B-Instruct found [here](https://github.com/QwenLM/Qwen2.5).


More details on model performance accross various devices, can be found [here](https://aihub.qualcomm.com/models/qwen2_7b_instruct_quantized).

### Model Details

- **Model Type:** Text generation
- **Model Stats:**
  - Input sequence length for Prompt Processor: 128
  - Context length: 4096
  - Number of parameters: 7.07B
  - Precision: w4a16 + w8a16 (few layers)
  - Num of key-value heads: 8
  - Information about the model parts: Prompt Processor and Token Generator are split into 5 parts each. Each corresponding Prompt Processor and Token Generator part share weights.
  - Prompt processor model size: 5.16 GB
  - Prompt processor input (part1): 128 tokens
  - Prompt processor output (part1): Embeddings output
  - Prompt processor input (other parts): 128 tokens + KVCache initialized with pad token
  - Prompt processor output (other parts): 128 output tokens + KVCache for token generator
  - Token generator model size: 5.16 GB
  - Token generator input (part1): 128 tokens
  - Token generator output (part1): Embeddings output
  - Token generator input (other parts): 1 input token + past KVCache
  - Token generator output (other parts): 1 output token + KVCache for next iteration
  - Use: Initiate conversation with prompt-processor and then token generator for subsequent iterations.
  - Minimum QNN SDK version required: 2.27.7
  - Supported languages: English, Chinese, German, French, Spanish, Portuguese, Italian, Dutch, Russian, Czech, Polish, Arabic, Persian, Hebrew, Turkish, Japanese, Korean, Vietnamese, Thai, Indonesian, Malay, Lao, Burmese, Cebuano, Khmer, Tagalog, Hindi, Bengali, Urdu.
  - 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).
  - 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)
|---|---|---|---|---|---|
| Qwen2-7B-Instruct | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 13.65 | 0.170593 - 5.458976 | -- | Use Export Script |

## Deploying Qwen2-7B-Instruct on-device

Please follow the [LLM on-device deployment]({genie_url}) tutorial.



## License
* The license for the original implementation of Qwen2-7B-Instruct can be found [here](https://huggingface.co/Qwen/Qwen2-7B-Instruct/blob/main/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://huggingface.co/Qwen/Qwen2-7B-Instruct/blob/main/LICENSE)



## References
* [Qwen2 Technical Report](https://arxiv.org/abs/2407.10671v1)
* [Source Model Implementation](https://github.com/QwenLM/Qwen2.5)



## 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:ai-hub-support@qti.qualcomm.com).

## 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