https://civitai.com/models/1309415
goto pyhon_embedded\lib\site-packages and delete all torch folders. Delete xformers folders. Do not delete torchsde folders. Then got back to python_embedded folder. right-click run terminal. Then run ".\python.exe -m pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128"
goto python_embedded\sagattention folder and run "..\python.exe -s -m pip install -e ."
This will take some time. After it is done. You can run run_nvidia_gpu-sageattention with your rtx 5000 card. This is what worked for me to use my 5080.
Requires CUDA 12.8.1
https://civitai.com/articles/12848
Started with the above setup on host for sage attention, not sure what parts are needed. Then:
git clone https://github.com/lllyasviel/FramePack
cd framepack
python -m venv venv
venv\Scripts\activate.bat
python.exe -m pip install --upgrade pip
pip install -r requirements.txt
pip uninstall torch torchvision torchaudio
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128
python.exe -s -m pip install triton-windows
pip install https://github.com/woct0rdho/SageAttention/releases/download/v2.1.1-windows/sageattention-2.1.1+cu128torch2.8.0-cp312-cp312-win_amd64.whl
pip install packaging
pip install ninja
set MAX_JOBS=4
pip install flash-attn --no-build-isolation
Then to run:
venv\Scripts\activate.bat
python.exe demo_gradio.py