library_name: transformers
license: mit
model_name: MBart-Urdu-Text-Summarization
pipeline_tag: summarization
tags:
- text-generation
- mbart
- nlp
- transformers
- text-generation-inference
author: Wali Muhammad Ahmad
private: false
gated: false
inference: true
mask_token: <mask>
widget_data:
  text: Enter your para here
transformers_info:
  auto_class: MBartForConditionalGeneration
  processor: AutoTokenizer
language:
- en
- ur
---

# Model Card

MBart-Urdu-Text-Summarization is a fine-tuned MBart model designed for summarizing Urdu text. It leverages the multilingual capabilities of MBart to generate concise and accurate summaries for Urdu paragraphs.

## Model Details

### Model Description

This model is based on the MBart architecture, which is a sequence-to-sequence model pre-trained on multilingual data. It has been fine-tuned specifically for Urdu text summarization tasks. The model is capable of understanding and generating text in both English and Urdu, making it suitable for multilingual applications.

### Model Sources [optional]

- **Repository:** [https://github.com/WaliMuhammadAhmad/UrduTextSummarizationUsingm-BART]
- **Paper [Multilingual Denoising Pre-training for Neural Machine Translation]:** [https://arxiv.org/abs/2001.08210]

## Uses

### Direct Use

This model can be used directly for Urdu text summarization tasks. It is suitable for applications such as news summarization, document summarization, and content generation.

### Downstream Use [optional]

The model can be fine-tuned for specific downstream tasks such as sentiment analysis, question answering, or machine translation for Urdu and English.

### Out-of-Scope Use

This model is not intended for generating biased, harmful, or misleading content. It should not be used for tasks outside of text summarization without proper fine-tuning and evaluation.

## Bias, Risks, and Limitations

- The model may generate biased or inappropriate content if the input text contains biases.
- It is trained on a specific dataset and may not generalize well to other domains or languages.
- The model's performance may degrade for very long input texts.

### Recommendations

Users should carefully evaluate the model's outputs for biases and appropriateness. Fine-tuning on domain-specific data is recommended for better performance in specialized applications.

## How to Get Started with the Model

Use the code below to get started with the model.

```python
from transformers import AutoTokenizer, MBartForConditionalGeneration

# Load the model and tokenizer
model_name = "ihatenlp/MBart-Urdu-Text-Summarization"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = MBartForConditionalGeneration.from_pretrained(model_name)

# Example input text
input_text = "Enter your Urdu paragraph here."

# Tokenize and generate summary
inputs = tokenizer(input_text, return_tensors="pt")
summary_ids = model.generate(inputs["input_ids"], max_length=50, num_beams=4, early_stopping=True)
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)

print("Summary:", summary)
```

## Environmental Impact

Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).

## Citation [optional]

**BibTeX:**

```bibtex
@misc{liu2020multilingualdenoisingpretrainingneural,
      title={Multilingual Denoising Pre-training for Neural Machine Translation}, 
      author={Yinhan Liu and Jiatao Gu and Naman Goyal and Xian Li and Sergey Edunov and Marjan Ghazvininejad and Mike Lewis and Luke Zettlemoyer},
      year={2020},
      eprint={2001.08210},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2001.08210}, 
}
```

## Model Card Authors [optional]

- **Wali Muhammad Ahmad**
- **Muhammad Labeeb Tariq**

## Model Card Contact

- **Email:** [wali.muhammad.ahmad@gmail.com]
- **Hugging Face Profile:** [Wali Muhammad Ahmad](https://huggingface.co/ihatenlp)