---
library_name: transformers
tags:
- text-generation-inference
license: cc-by-4.0
datasets:
- AhmadMustafa/Urdu-Instruct-News-Article-Generation
language:
- ur
---

# Model Card for Model ID

This is Instruct Fine-tuned Version of [MobiLlama](https://arxiv.org/abs/2402.16840) Fine-tuned on [Instruct Urdu Article Generation Dataset](https://huggingface.co/datasets/AhmadMustafa/Urdu-Instruct-News-Article-Generation). Instruct Urdu Article Generation Dataset was released under [AYA Collections](https://arxiv.org/abs/2402.06619) by [Cohere for AI](cohere.for.ai)

This model is finetuned for 8500 steps for generating articles in Urdu Language.


Fine-Tuning Statistics:
![image/png](https://cdn-uploads.huggingface.co/production/uploads/6246908d8031dcfa9ef6d80b/Y9t_6KZ8Uloe0N16yqTPk.png)


### Model Description

<!-- Provide a longer summary of what this model is. -->

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

- **Developed by:** [Ahmad Mustafa Anis]
- **Language(s) (NLP):** [Urdu]
- **License:** [CC by 4.0]
- **Finetuned from model [optional]:** [MBZUAI/MobiLlama-05B]

### Model Sources [optional]

<!-- Provide the basic links for the model. -->

- **Repository:** [https://github.com/mbzuai-oryx/MobiLlama?tab=readme-ov-file]
- **Paper [optional]:** [https://arxiv.org/abs/2402.16840]

## Uses
This model is intended to use on mobile devices for generating articles in Urdu Language.
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

## Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->
This model may contain biases and limitations that are present in LLMs and I have not accounted for them.

## How to Get Started with the Model

Use the code below to get started with the model.

```python3

model = AutoModelForCausalLM.from_pretrained("AhmadMustafa/MobiLLama-Urdu-Article-Generation", trust_remote_code=True).to(device)
tokenizer = AutoTokenizer.from_pretrained("MBZUAI/MobiLlama-05B")

example = {'inputs': """ اس دی گی ایک خبر سے متعلق ایک مضمون لکھیں۔
خبر: سشانت سنگھ کیس بھارتی سپریم کورٹ نے فریقین سے مفصل جواب طلب کرلیا""",
}

example = f"### Instruction: {example['inputs']}\n ### Completion: "
inputs = tokenizer.encode(f"{example}", return_tensors="pt").to(device)

outputs = model.generate(inputs, max_new_tokens=512)

print(tokenizer.decode(outputs[0]))
```

Please note that I have used <|EOS|> in the end of each example so you can use that as ending token to control generation.
## Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

@misc{thawakar2024mobillama,
      title={MobiLlama: Towards Accurate and Lightweight Fully Transparent GPT}, 
      author={Omkar Thawakar and Ashmal Vayani and Salman Khan and Hisham Cholakkal and Rao Muhammad Anwer and Michael Felsberg and Timothy Baldwin and Eric P. Xing and Fahad Shahbaz Khan},
      year={2024},
      eprint={2402.16840},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
} 

## Model Card Authors [optional]

- Name: Ahmad Mustafa Anis
- Email: ahmadanis5050@gmail.com