Comment by lz400

Unfortunately uv is usually insufficient for certain ML deployments in Python. It's a real pain to install pytorch/CUDA with all the necessary drivers and C++ dependencies so people tend to fall back to conda.

Any modern tips / life hacks for this situation?