Running into a bit of a issue here and hoping someone with more experience can assist or shed some light on a solution with a 5090 to get working for training.
Clean install of Windows 11. Running python-3.10.9 with Cuda 12.8 installed minus Nsight Compute because it keeps failing the install if that is checked in the Cuda download exe.
When trying to run: “pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu128” throws the error “Could not find a version that satisfies the requirement torchaudio (from versions: none)” and “ERROR: No matching distribution found for torchaudio”
I had to downgrade pip to 20.0.2 to even get the above to even consider downloading it. I am trying to get LoRA_Easy_Training_Scripts working again, but it throws a error of “NVIDIA GeForce RTX 5090 with CUDA capability sm_120 is not compatible with the current PyTorch installation.The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_61 sm_70 sm_75 sm_80 sm_86 sm_90.” Which all points to getting the nightly build installed to get it to work I am assuming.
Any help would be greatly appreciated since I am not the most knowledgeable on what workaround I should be trying