Dallah Llama

🚀 Dallah: A Dialect-Aware Multimodal Large Language Model for Arabic

Dallah is an advanced multimodal large language model (MLLM) designed specifically for the Arabic language. It focuses on understanding and generating content in various Arabic dialects. Built on the LLaVA framework and powered by the LLaMA - 2 architecture, Dallah combines textual and visual data to enable comprehensive multimodal interactions.

🚀 Quick Start

This section is not provided in the original README, so it is skipped.

✨ Features

• Tailored for the Arabic language, with support for Modern Standard Arabic (MSA) and six major Arabic dialects.
• Integrates textual and visual data for comprehensive multimodal interactions.
• Demonstrates state - of - the - art performance in Arabic MLLMs, excelling in both MSA and dialectal Arabic benchmarks.
• Capable of handling complex multimodal interactions involving text and images.

📦 Installation

This section is not provided in the original README, so it is skipped.

💻 Usage Examples

This section is not provided in the original README, so it is skipped.

📚 Documentation

Model Details

Training Data

Dallah was fine - tuned on a diverse dataset that includes both textual and visual information:

• Textual Data: It contains MSA and six prominent Arabic dialects, ensuring the model's proficiency across different regional linguistic variations.
• Visual Data: Composed of image - text pairs, allowing the model to process and generate content that integrates both modalities.

Performance

Dallah shows state - of - the - art performance in Arabic MLLMs:

• It performs excellently in both MSA and dialectal Arabic benchmarks.
• It can effectively handle complex multimodal interactions involving textual and visual elements.

Applications

Dallah’s multimodal and dialect - aware capabilities make it suitable for a variety of applications:

• Multilingual Chatbots: Enhance user interactions by understanding and responding in specific Arabic dialects.
• Content Creation: Assist in generating culturally and linguistically appropriate content for diverse Arabic - speaking audiences.
• Educational Tools: Support language learning by providing examples and explanations in various dialects.
• Cultural Preservation: Document and promote the use of different Arabic dialects on digital platforms.

🔧 Technical Details

This section is not provided in the original README, so it is skipped.

📄 License

This section is not provided in the original README, so it is skipped.

📚 Citation

If you use Dallah in your research or applications, please cite the following paper:

@inproceedings{alwajih2024dallah,
  title={Dallah: A Dialect-Aware Multimodal Large Language Model for Arabic},
  author={Alwajih, Fakhraddin and Bhatia, Gagan and Abdul-Mageed, Muhammad},
  booktitle={Proceedings of The Second Arabic Natural Language Processing Conference},
  pages={320--336},
  year={2024},
  address={Bangkok, Thailand},
  publisher={Association for Computational Linguistics},
  url={https://aclanthology.org/2024.arabicnlp-1.27}
}