Torch.cuda.is_available() is always False for 12.1

windows
1

Hi everyone,

I am having troubles setting up CUDA for my windows computer. I am unsure what the issue is as I tried to debug the problem with some of the solutions on the forum.

Blockquote
torch.utils.collect_env

image
image912×348 8.76 KB

Blockquote

any help is appreciated

2

nvidia-smi
image

3

install with the following code.

conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia
conda install cuda -c nvidia

4

RESOLVED -

conda activate cuda_venv
conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia
conda install -c anaconda cudatoolkit
pip install jupyter
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
jupyter notebook

My problem was that there was there was another kernel I created while following a tutorial on setup and that still has a false value. I swapped back to the python3 ipynkernel.