Fine-Tuned Medical AI Assistant

Meet Galenos

Fine-tuning Microsoft Phi-3-mini on medical dialogue data using QLoRA, PEFT, and TRL. Experience structured medical assistance powered by state-of-the-art LLM techniques.

🧬

QLoRA Fine-Tuning

Efficient 4-bit quantization with LoRA adapters enables training on consumer hardware without sacrificing performance.

📋

Structured Responses

Galenos provides organized medical guidance with symptoms summary, differential diagnosis, red flags, and next steps.

Production Ready

Deployed on Vercel with Groq API integration for lightning-fast inference and seamless scalability.

Technology Stack

🤖
Phi-3-mini-4k
Base Model
⚙️
bitsandbytes
4-bit Quantization
🔧
PEFT (LoRA)
Adapter Training
📚
TRL SFTTrainer
Supervised Fine-Tuning
Groq API
Fast Inference
Next.js 14
Web Framework
🎨
Tailwind CSS
Styling
🚀
Vercel
Deployment

Ready to Experience It?

Try the live demo and see how fine-tuned medical AI can provide structured, helpful guidance.

Launch Demo →