Torch.cuda.is_available() : False CUDA:12.1 [Solved]

torch.utils.collect_env :

Can anyone please help.

nvidia-smi :

nvcc --version :

I’ve managed to solved this on my own.

I was initially trying to install using conda.
I uninstalled pytorch and tried reinstalling using pip and this time everything worked perfectly.

This post may be closed now.

1 Like

Hi, Mayur, thanks for sharing your experience. I met the same problem here. But I cannot install pytorch==1.12.1 with cuda==12.1 by using pip. What you installed is still pytorch 1.12.1 or pytorch 2.1? If possible, could you please share your installation command? Thank you so much!