My CUDA version is 12.1, which version of pytorcn should I choose, the latest 2.0, or should I choose to downgrade the driver and lower the CUDA version?
The PyTorch binaries ship with their own CUDA runtime and math libs. Your local CUDA toolkit will be used if you build PyTorch from source or custom extensions. You can thus install the current release with the bundled CUDA 11.7 or 11.8 libs.
Thanks, I’m a beginner and don’t know how to install the current release with the bundled CUDA 11.7 or 11.8 libs, can you tell me what command to use?
Select the desired CUDA version from the install matrix here, copy/paste the command into your terminal, and execute it.
I would also recommend to either use a new and empty virtual environment or to make sure no other PyTorch packages are installed in the current environment.