app.py

import gradio as gr
import torch
from cadfusion.models import CADFusionModel

# Load model (from HF weights + GitHub code)
print("Loading CADFusion model...")
device = "cuda" if torch.cuda.is_available() else "cpu"

model = CADFusionModel.from_pretrained("microsoft/CADFusion")
model = model.to(device)
model.eval()

def generate(prompt):
    """Run CADFusion inference on user prompt"""
    with torch.no_grad():
        output = model.generate(
            prompt,
            max_new_tokens=256,
            temperature=0.7,
            top_p=0.9,
        )
    return output

# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("## 🏗️ CADFusion Demo\nEnter a CAD prompt below:")
    inp = gr.Textbox(label="Your CAD prompt")
    out = gr.Textbox(label="Model Output")

    btn = gr.Button("Generate")
    btn.click(fn=generate, inputs=inp, outputs=out)

if __name__ == "__main__":
    demo.launch()