# Update system
sudo apt update && sudo apt upgrade -y

# Install Python and basic tools
sudo apt install -y python3 python3-pip python3-venv git

# Create project directory
mkdir -p ~/lancet3 && cd ~/lancet3

# Create Python virtual environment
python3 -m venv venv
source venv/bin/activate

# Upgrade pip
pip install --upgrade pip

# Install inference framework
pip install transformers torch accelerate

# Install Web UI framework (optional, for building a chat interface)
pip install gradio fastapi uvicorn

# Or use Docker deployment (recommended)
# Install Docker
curl -fsSL https://get.docker.com | sh
sudo usermod -aG docker $USER

# Clone the repository
git clone https://github.com/Tencent/Huatuo-Llama-Med-Chinese.git
cd Huatuo-Llama-Med-Chinese

# Or load directly from Hugging Face
python3 << 'EOF'
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load Lancet-3 model (7B variant example)
model_name = "Tencent/Lancet-3-7B"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto",
    trust_remote_code=True
)

print("✅ Model loaded successfully!")
print(f"Model path: {model_name}")
print(f"Parameters: 7B")
EOF

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull the Lancet-3 medical model
ollama pull lancet-3

# Start API service
ollama serve

# Test conversation
curl http://localhost:11434/api/generate -d '{
  "model": "lancet-3",
  "prompt": "Patient has had persistent fever for three days, with cough and chest tightness. What could it be?",
  "stream": false
}'

# Create Docker Compose file
cat > docker-compose.yml << 'EOF'
version: '3.8'
services:
  lancet3:
    image: ghcr.io/tencent/lancet-3:7b
    container_name: lancet3-medical-ai
    ports:
      - "8080:8080"
    environment:
      - MODEL_NAME=Lancet-3-7B
      - PORT=8080
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    volumes:
      - ./models:/app/models
      - ./data:/app/data
    restart: unless-stopped

  # Optional: Web UI companion
  chat-ui:
    image: gradio/gradio
    ports:
      - "7860:7860"
    depends_on:
      - lancet3
    environment:
      - API_URL=http://lancet3:8080
EOF

# Start services
docker compose up -d

# app.py — Medical chat interface based on Gradio
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "Tencent/Lancet-3-7B"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto",
    trust_remote_code=True
)

def medical_chat(user_input, history):
    messages = []
    for h in history:
        messages.append({"role": "user", "content": h[0]})
        messages.append({"role": "assistant", "content": h[1]})
    messages.append({"role": "user", "content": user_input})
    
    prompt = tokenizer.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    
    return response

with gr.Blocks(title="Lancet-3 Medical AI Assistant") as demo:
    gr.Markdown("# 🏥 Lancet-3 Medical AI Assistant\n> Privately deployed medical professional AI, data stays on-premise")
    
    chatbot = gr.Chatbot(height=400)
    msg = gr.Textbox(placeholder="Enter your medical inquiry...", lines=2)
    
    with gr.Row():
        clear = gr.Button("🗑️ Clear")
        submit = gr.Button("📤 Send")
    
    def respond(message, history):
        if not message.strip():
            return history, ""
        response = medical_chat(message, history)
        return history + [[message, response]], ""
    
    msg.submit(respond, [msg, chatbot], [chatbot, msg])
    submit.click(respond, [msg, chatbot], [chatbot, msg])
    clear.click(lambda: None, None, chatbot, queue=False)

demo.queue()
demo.launch(server_name="0.0.0.0", server_port=7860)

python3 app.py
# Visit http://your-vps-ip:7860

# Install Caddy (automatic HTTPS)
sudo apt install -y debian-keyring debian-archive-keyring apt-transport-https
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/gpg.key' | sudo gpg --dearmor -o /usr/share/keyrings/caddy-stable-archive-keyring.gpg
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/debian.deb.txt' | sudo tee /etc/apt/sources.list.d/caddy-stable.list
sudo apt update
sudo apt install -y caddy

# Configure Caddyfile
sudo tee /etc/caddy/Caddyfile << 'EOF'
medical.yourdomain.com {
    reverse_proxy localhost:7860
    
    header {
        X-Content-Type-Options nosniff
        X-Frame-Options DENY
    }
}
EOF

sudo systemctl reload caddy

# 1. Configure firewall, open only necessary ports
sudo ufw allow 443/tcp
sudo ufw allow 22/tcp
sudo ufw enable

# 2. Configure fail2ban to prevent brute force attacks
sudo apt install -y fail2ban
sudo systemctl enable fail2ban

# 3. Encrypt model files at rest
apt install -y ecryptfs-utils
ecryptfs-setup-private

# 4. Regularly update models and dependencies
pip install --upgrade transformers torch