-
I installed whisper and pytorch via pip. It run super slow and torch.cuda.is_available() showed false. Could not get that to show true via any help using pip. I uninstalled it and re installed via conda. Now it shows true but Anaconda seems only to run in its own shell where it can't find whisper. Below are my data. My CUDA is 1.18 but torch seems to have only latest version 1.16. However I doubt that this is the issue as it works when installing with conda. How do I get this to run ? nvidia-smi CUda 11.8 C:\Windows\system32>python
|
Beta Was this translation helpful? Give feedback.
Replies: 4 comments 8 replies
-
This worked for me.
|
Beta Was this translation helpful? Give feedback.
-
thanks it finally worked by using cu116, I was always trying to use a
higher version
…On Tue, Oct 25, 2022 at 6:53 PM J ***@***.***> wrote:
This worked for me.
pip uninstall torch
pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
I hit this same issue when I tried setting up Whisper on a Windows machine after using it on Linux for a long time. Took me hours to figure out the fix because I didn't know about this discussion area. It sure would be nice if this could be fixed to install a CUDA compatible version of pytorch automatically! |
Beta Was this translation helpful? Give feedback.
-
Isn't there some paramater we can use that forces the system to use cuda when its already working with other AIs without a problem? something like: model.transcribe("audio.mp3", gpu) |
Beta Was this translation helpful? Give feedback.
This worked for me.