GATE AraBert V1

🚀 GATE-AraBert-V1

This is GATE | General Arabic Text Embedding trained using SentenceTransformers in a multi-task setup. The system trains on the AllNLI and on the STS dataset. It is described in detail in the paper GATE: General Arabic Text Embedding for Enhanced Semantic Textual Similarity with Hybrid Loss Training. This model can effectively handle Arabic text embedding tasks, providing high - quality semantic representation for Arabic text.

Project page: https://huggingface.co/collections/Omartificial-Intelligence-Space/arabic-matryoshka-embedding-models-666f764d3b570f44d7f77d4e

🚀 Quick Start

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("Omartificial-Intelligence-Space/GATE-AraBert-v1")
# Run inference
sentences = [
    'الكلب البني مستلقي على جانبه على سجادة بيج، مع جسم أخضر في المقدمة.',
    'لقد مات الكلب',
    'شخص طويل القامة',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

✨ Features

• Multi - task Training: Trained on both the AllNLI and STS datasets, enabling it to handle a variety of semantic tasks.
• High - quality Embedding: Can generate high - dimensional and semantically rich Arabic text embeddings.
• Cosine Similarity: Uses cosine similarity as the similarity function to measure the semantic similarity between texts.

📦 Installation

To use this model, you need to install the Sentence Transformers library. You can install it using the following command:

pip install -U sentence-transformers

💻 Usage Examples

Basic Usage

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("Omartificial-Intelligence-Space/GATE-AraBert-v1")
# Run inference
sentences = [
    'الكلب البني مستلقي على جانبه على سجادة بيج، مع جسم أخضر في المقدمة.',
    'لقد مات الكلب',
    'شخص طويل القامة',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

📚 Documentation

Model Details

Model Description

Evaluation

Metrics

Semantic Similarity (sts - dev)

• Dataset: sts - dev
• Evaluated with EmbeddingSimilarityEvaluator

Semantic Similarity (sts - test)

• Dataset: sts - test
• Evaluated with EmbeddingSimilarityEvaluator

🔧 Technical Details

The model is based on the Sentence Transformers framework and uses a multi - task training approach. It is trained on the AllNLI and STS datasets to learn semantic information from different aspects. The base model is [Omartificial - Intelligence - Space/Arabic - Triplet - Matryoshka - V2](https://huggingface.co/Omartificial - Intelligence - Space/Arabic - Triplet - Matryoshka - V2), which provides a solid foundation for the model's performance.

📄 License

This project is licensed under the apache - 2.0 license.

👏 Acknowledgments

The author would like to thank Prince Sultan University for their invaluable support in this project. Their contributions and resources have been instrumental in the development and fine - tuning of these models.

📖 Citation

If you use the GATE, please cite it as follows:

@misc{nacar2025GATE,
      title={GATE: General Arabic Text Embedding for Enhanced Semantic Textual Similarity with Hybrid Loss Training}, 
      author={Omer Nacar, Anis Koubaa, Serry Taiseer Sibaee and Lahouari Ghouti},
      year={2025},
      note={Submitted to COLING 2025},
      url={https://huggingface.co/Omartificial-Intelligence-Space/GATE-AraBert-v1},
}