search

etched-ai / open-oasis

Inference script for Oasis 500M
MIT License
1.51k stars 126 forks source link

assert torch.cuda.is_available() AssertionError #31

Open xxoh111 opened 1 week ago

xxoh111 commented 1 week ago

Traceback (most recent call last): File "D:\soft\2 osas\open-oasis-master\open-oasis-master\generate.py", line 19, in assert torch.cuda.is_available() ^^^^^^^^^^^^^^^^^^^^^^^^^ AssertionError

I tried downloading from https://developer.nvidia.com/cuda-downloads to resolve the error, but it didn't help. please tell me who knows how to fix this. I don’t know what “cuda” is and I apologize if the question is stupid or it has already been asked, I couldn’t find the answer even on the Internet. I use windows 10

(sorry for errors in the text, I used Google translator)

VoidGitHub commented 1 week ago

CUDA only works with NVIDIA graphics cards

https://developer.nvidia.com/cuda-downloads

xxoh111 commented 1 week ago

CUDA only works with NVIDIA graphics cards

https://developer.nvidia.com/cuda-downloads

This is probably a problem with my computer as I downloaded this and it didn't help me.

VoidGitHub commented 1 week ago

What video card do you have?

matth3wmajf commented 1 week ago

You must be using an NVIDIA graphics card, however you may likely be able to use this with OpenCL allowing you to use a different GPU architecture other than NVIDIA. Or maybe you may not have the NVIDIA drivers properly set up. What graphics card shows up for you on Device Manager (if on windows) or PCIe device listing (lspci on UNIX-like systems)?

xxoh111 commented 1 week ago

What video card do you have?

NVIDIA GeForce GTX 1650 Ti

xxoh111 commented 1 week ago

You must be using an NVIDIA graphics card, however you may likely be able to use this with OpenCL allowing you to use a different GPU architecture other than NVIDIA. Or maybe you may not have the NVIDIA drivers properly set up. What graphics card shows up for you on Device Manager (if on windows) or PCIe device listing (lspci on UNIX-like systems)?

NVIDIA GeForce GTX 1650 Ti

OraCatQAQ commented 1 week ago

@xxoh111 I had the same problem as you, but I solved it Try these

  1. Make sure you have the driver for your NVIDIA graphics card, then download the CUDA Toolkit
  2. Run nvcc --version from the command line to verify that CUDA is installed correctly
  3. Uninstall the current PyTorch version pip uninstall torch torchvision torchaudio
  4. Reinstall pytorch pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121
  5. check again