Closed welldawell closed 2 months ago
You will need to download the respective CUDA software, i have found 11.8 works and then use the PyTorch 11.8 as well installing after the initial install. Keep with Python 3.11.x. This worked for me today.
Thanks, but still the same error on my side. This is what I did:
I uninstalled CUDA 12.1 and installed the CUDA Toolkit 11.8 by downloading it from https://developer.nvidia.com/cuda-11-8-0-download-archive?target_os=Windows&target_arch=x86_64&target_version=11&target_type=exe_local for my installed Windows 11 OS.
I opened the cmd and uninstalled insanely-fast-whisper by using
pipx uninstall insanely-fast-whisper
I installed it again using
pipx install insanely-fast-whisper --force --pip-args="--ignore-requires-python"
I uninstalled torch by using
pip3 uninstall torch torchvision torchaudio
I installed torch again by using
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
As I got the same error message when I executed
insanely-fast-whisper --file-name untitled.wav
I tried again to uninstall torch and install it again using a modified fix for the error from the GitHub page:
python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
Still I get the error message "AssertionError: Torch not compiled with CUDA enabled".
What am I doing wrong? Thanks again.
Try uninstalling with pipx pipx uninstall insanely-fast-whisper
then installing with pip instead pip install insanely-fast-whisper --ignore-requires-python
that worked for me
This finally works also for me, thanks!
I followed the instructions on in the readme on https://github.com/Vaibhavs10/insanely-fast-whisper to install insanely-fast-whisper using a command line in Windows 11 and
pipx install insanely-fast-whisper --force --pip-args="--ignore-requires-python"
(I'm using Python 3.11.4).I also used the fix for the error message of the title after it appeared the first time:
python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
But when re-executing the command
insanely-fast-whisper --file-name untitled.wav
the following output is still shown: `Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.