Can we get a normal venv setup with a requirements.txt?

Trying the UV steps and no matter what order I try them in the run_gradio.py gives this error
Warning: You are loading the medium model without a GPU. This model is not designed to run on cpu
If you just provided a basic requirements.txt like every other setup then the user creates a venv, pip installs requirements.txt and then their specific version of pytorch.

What am I doing wrong here? This is a batch file to try and automate the install on Windows.

@echo off

echo *** %time% *** Deleting stable-audio-3 directory if it exists
if exist stable-audio-3\. rd /S /Q stable-audio-3

echo *** %time% *** Cloning repository
git clone https://github.com/Stability-AI/stable-audio-3
cd stable-audio-3

echo *** %time% *** Creating venv
python -m venv .venv

echo *** %time% *** Activating venv
call .venv\scripts\activate.bat

echo *** %time% *** Installing requirements
python.exe -m pip install --upgrade pip
pip install uv
uv pip install torch==2.7.1 torchaudio==2.7.1 --index-url https://download.pytorch.org/whl/cu118
uv sync --no-install-package torch --no-install-package torchaudio
uv pip install https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.16/flash_attn-2.6.3+cu126torch2.7-cp310-cp310-linux_x86_64.whl
uv sync --extra ui --extra lora

call .venv\scripts\deactivate.bat
cd ..
echo *** %time% *** Finished Stable Audio Tools install
echo.
echo Check the stats for any errors.  Do not assume it worked.
pause

Start with run_gradio.py and the Windows batch script in the issue, then reproduce the UV-based setup in a clean Python virtual environment. Identify which installation steps are required for the project and define a requirements.txt-based path that handles the documented PyTorch choice. Done means the setup instructions work for the intended environment without the reported installation confusion.