KurMed-Triage-XLMRoBERTa

Medical specialty classification model for Kurdish and Persian speakers, fine-tuned on KurMed-Triage dataset.

Model Performance

Metric	Score
Accuracy	87.0%
F1 Score (weighted)	0.871

Usage

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="alanjafari/KurMed-Triage-XLMRoBERTa"
)

result = classifier("درد قفسه سینه دارم و نفس کشیدنم سخته")
print(result)

Classes (10 Specialties)

cardiology, dermatology, emergency, ent, gynecology, internal_medicine, neurology, orthopedics, pediatrics, psychiatry

Training Details

• Base model: xlm-roberta-base
• Dataset: 2,000 samples (1,700 train / 300 validation)
• Epochs: 5
• Input: Persian + English patient text
• Hardware: Google Colab T4 GPU

Dataset

alanjafari/KurMed-Triage

Author

Alan Jafari — AI Specialist & Founder of EONE Team