🇹🇷 Turkish Medical Triage LLaMA3 - GGUF

🔗 Model Hugging Face Link: https://huggingface.co/dousery/turkish-medical-triage-llama3-gguf

An open-source Turkish-language medical triage and diagnosis model, fine-tuned on synthetic emergency medical data. Built on Meta's LLaMA 3 and quantized to GGUF format for efficient offline inference.

⚕️ This model predicts both the urgency level of a Turkish medical complaint and provides a clear response message, emulating how a virtual assistant might reply in a triage scenario.

---

🧠 Model Capabilities

• 🔍 Input: Free-form Turkish medical complaint sentence
• 🎯 Output: Natural-language response that includes:
  • Triage level (e.g. ACIL, ORTA, NORMAL)
  • Possible diagnosis
  • Clear, directive language in Turkish

📌 Example

{
  "input": "nefes alamıyorum, göğsümde batıcı ağrı var ve terliyorum",
  "response": "ACIL DURUM: Anında 112'yi arayın. Kalp krizi belirtileri gösteriyorsunuz. Derhal hastaneye gitmeniz gerekiyor."
}

---

🚀 Use Cases

• Emergency virtual triage agents
• Turkish healthcare chatbots
• Simulation for training emergency responders
• Research in multilingual healthcare LLMs

---

🔧 Inference with GGUF

You can run this model using:

• llama.cpp
• text-generation-webui
• LM Studio

🖥️ Example with llama.cpp:

./main -m turkish-medical-triage-llama3.Q4_K_M.gguf -p "başım çok dönüyor, ayakta duramıyorum ve kusuyorum"

Expected Output:

ORTA DERECE RİSK: Muhtemelen bir iç kulak enfeksiyonu veya mide rahatsızlığı yaşıyor olabilirsiniz. En kısa sürede bir sağlık kuruluşuna başvurmanız önerilir.

---

📊 Dataset Summary

Each training sample includes:

• A Turkish-language medical complaint sentence
• Annotated symptoms
• A possible diagnosis
• A natural-language response combining triage + advice

---

📁 Files

• README.md
• *.gguf – Quantized model files
• tokenizer.model

---

✍️ Author

• Dousery
  🧪 Turkish NLP / AI researcher focused on public-benefit applications

---

📜 License

MIT License

⚠️ Disclaimer: This model is not intended for real-life clinical use. Do not use for medical decisions without professional supervision.

---

💬 Suggestions, feedback, and contributions are welcome!