app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load CAD-Coder model from Hugging Face
MODEL_NAME = "CADCODER/CAD-Coder"

tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    trust_remote_code=True,
    torch_dtype=torch.float16,
    device_map="auto"
)

def generate_code(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=512,
            do_sample=True,
            temperature=0.7,
            top_p=0.9
        )
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Gradio UI
demo = gr.Interface(
    fn=generate_code,
    inputs=gr.Textbox(lines=5, placeholder="Enter your CAD design prompt..."),
    outputs="text",
    title="CAD-Coder Inference",
    description="Generate CAD code from natural language using CAD-Coder."
)

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