README.md

---
library_name: transformers
language: ar
pipeline_tag: text-classification
tags:
- text-classification
- intent-classification
- marbert
- egyptian-arabic
- nlu
- e-commerce
- customer-service
license: apache-2.0
---

# 🌍 MARBERT for Egyptian Dialect Intent Classification (syplyd-marbert-v1)

This is a fine-tuned version of [UBC-NLP/MARBERTv2](https://huggingface.co/UBC-NLP/MARBERTv2), specifically adapted for **intent classification** in **Egyptian Colloquial Arabic**, with a primary focus on **e-commerce** and **customer service** scenarios.

It enables accurate understanding of user queries in dialectal Arabic, empowering applications like chatbots, support assistants, and ticket routing systems.

---

## 🧠 Model Details

- **Model Type**: `bert-for-sequence-classification`
- **Base Model**: [UBC-NLP/MARBERTv2](https://huggingface.co/UBC-NLP/MARBERTv2)
- **Language**: Arabic (Egyptian dialect)
- **Developer**: Shaza Aly
- **License**: Apache 2.0
- **Repository**: [https://huggingface.co/ShazaAly/syplyd-marbert-1](https://huggingface.co/ShazaAly/syplyd-marbert-1)

---

## 🚀 Usage

This model can be used directly with the Hugging Face `transformers` library:

```python
from transformers import pipeline

classifier = pipeline("text-classification", model="ShazaAly/syplyd-marbert-1")

# Example 1
text_1 = "عايز أعرف الأوردر بتاعي هيوصل امتى؟"
print(classifier(text_1))
# Output: [{'label': 'track_order_status', 'score': ...}]

# Example 2
text_2 = "المنتج ده غالي، فيه بديل أرخص؟"
print(classifier(text_2))
# Output: [{'label': 'product_alternatives', 'score': ...}]