xlm-roberta-base-tarifit-university-v0.8

---

language:

• ber # Tarifit / Tamazight
• es # Spanish
• ca # Catalan license: apache-2.0 tags:
• nlp
• text-classification
• xlm-roberta
• tamazight
• university-intents datasets:
• custom-tarifit-university-v0.8 metrics:
• accuracy

---

XLM-RoBERTa for Tarifit University Intent Classification (v0.8)

📌 Project Overview

This model is a specialized Intent Classifier designed for academic and university environments. It has been fine-tuned to recognize and categorize student inquiries in Tarifit (Riffian Berber), as well as Spanish and Catalan.

As a project led by a Linguistics & Philology expert, this model aims to bridge the digital gap for underrepresented languages (Amazigh) by providing high-quality Natural Language Processing (NLP) tools for administrative and educational contexts.

📊 Model Details

• Base Model: xlm-roberta-base (Cross-lingual Language Model)
• Version: 0.8
• Language Coverage: Trilingual (Tarifit, Spanish, Catalan)
• Classes: 5 Intent Categories:
  1. Enrollment / Registration
  2. Scholarships & Financial Aid
  3. Schedules & Calendars
  4. Exams & Grading
  5. General Inquiries

🧪 Training Setup

• Dataset Size: 1,000 manually curated and linguistically verified samples.
• Training Epochs: 15
• Optimizer: AdamW
• Framework: Hugging Face Transformers & PyTorch

🛠️ Intended Use

This model is ideal for:

• Building Chatbots for university administration in North Africa or regions with Amazigh-speaking students.
• Automated Email Routing for student support services.
• Research in Low-Resource Language fine-tuning.

👤 About the Author

Developed by a Linguist specializing in Computational Linguistics and Amazigh Studies. This project demonstrates the intersection of traditional philology and state-of-the-art Deep Learning to preserve and empower the Riffian language in the digital age.

---

Disclaimer: This is a research model (v0.8). Performance may vary with highly informal dialects or slang.