import os

from inference.interface import PenguinVLQwen3GradioInterface
from inference.server import PenguinVLQwen3DirectClient
from inference.server.direct_client import ensure_flash_attn_installed


def main():
    ensure_flash_attn_installed()
    model_client = PenguinVLQwen3DirectClient(
        model_path=os.getenv("MODEL_PATH", "tencent/Penguin-VL-8B"),
    )
    interface = PenguinVLQwen3GradioInterface(
        model_client,
        example_dir=os.getenv("EXAMPLE_DIR", "./assets/inputs"),
        server_name=os.getenv("GRADIO_SERVER_NAME", "0.0.0.0"),
        server_port=int(os.getenv("PORT", "7860")),
    )
    interface.launch()


if __name__ == "__main__":
    main()