Troubleshooting

NVIDIA driver / CUDA mismatch with PyTorch

If you see errors about the NVIDIA driver being too old for the installed PyTorch build, align driver and torch builds using the official PyTorch install matrix.

Example pin used in this project’s README:

pip uninstall torch
pip install torch==2.9.1

Adjust the version to match your CUDA / driver stack.

SpaceMouse (Linux udev)

Add a rule (example for SpaceMouse Compact) to /etc/udev/rules.d/99-spacemouse.rules:

KERNEL=="hidraw*", ATTRS{idVendor}=="256f", ATTRS{idProduct}=="c635", MODE="0666", GROUP="plugdev"
SUBSYSTEM=="usb", ATTRS{idVendor}=="256f", ATTRS{idProduct}=="c635", MODE="0666", GROUP="plugdev"

Reload rules:

sudo udevadm control --reload-rules

pyspacemouse: pyspacemouse==1.1.4 is known to work; 2.0.0 has been reported incompatible.

MediaPipe

Pin mediapipe==0.10.21 if you hit errors related to solutions on newer releases.

Still stuck?

Open an issue on the OopsieVerse repository with your OS, GPU, driver version, and the exact command plus traceback.