Hi, I have Windows 11 and I never installed Cuda 12.6. I have a clean Cuda 11.8 and CuDNN install. When running: conda install pytorch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 pytorch-cuda=11.7 -c pytorch -c nvidia, it installs Cuda 12.6 libraries instead Cuda 11.8 libraries. This causes code to get error like RuntimeError: Library cublas64_12.dll is not found or cannot be loaded. The Cuda 12 libraries are: cuda-cccl, cuda-cccl_win-64 and cuda-version 12.6.37. This happens when I run any combination of versions for the command listed above, even when I do not specify the version. If someone could tell me what I am doing wrong I would appreciate it.
Are you using conda-forge as a supported channel? If so, could you remove it and use the default main one instead?
Thank you, but I tried conda-forge, I tried pip with pip install torch==2.0.0+cu118 torchvision==0.15.0+cu118 torchaudio==2.0.0+cu118 -f https://download.pytorch.org/whl/torch_stable.html
and I still have the same error since all of these commands are pulling 12.6 libraries.