app.py

import gradio as gr
from huggingface_hub import InferenceClient

MODEL = "epfl-llm/meditron-7b"

def chat(message, history, hf_token):
    if not hf_token: return "Please log in"
    
    client = InferenceClient(token=hf_token.token, model=MODEL)
    response = client.text_generation(f"Medical tutor: {message}", max_new_tokens=150)
    return response

gr.ChatInterface(
    chat,
    title="🩺 Medical Tutor",
    examples=["Explain how vaccines work", "What is DNA?"]
).launch(server_port=7860)