import gradio as gr

# Usa a API gratuita do Hugging Face para rodar o modelo sem consumir sua RAM
demo = gr.load("models/Qwen/Qwen2.5-0.5B-Instruct")

demo.launch()