search

nerdyrodent / VQGAN-CLIP

Just playing with getting VQGAN+CLIP running locally, rather than having to use colab.
Other
2.61k stars 427 forks source link

which CUDA version is required for pytorch here? #8

Closed christiansievers closed 3 years ago

christiansievers commented 3 years ago

I'm getting UserWarning: CUDA initialization: CUDA driver initialization failed, you might not have a CUDA gpu. (Triggered internally at /pytorch/c10/cuda/CUDAFunctions.cpp:115.). I do have a GPU, but I'm using CUDA version 8 (it's a shared lab machine).

Is the old CUDA version why I get the above error? Any way to fix this, apart from setting up a brand new system?

nerdyrodent commented 3 years ago

I'm using CUDA 11 as it's a requirement for the RTX 30x series: pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html Not actually tested any versions prior, but CUDA 10 should also be OK.

christiansievers commented 3 years ago

Right. Thanks. I guess it's time to set everything up from scratch again.

Just to check, in your instructions you mention that CLIP and taming-transformers also need to be present. Is just git cloning them enough, or do they also need to be installed, following the instructions on their respective pages?

nerdyrodent commented 3 years ago

I have OpenAI CLIP and taming-transformers available like this:

git clone https://github.com/openai/CLIP
git clone https://github.com/CompVis/taming-transformers.git

CLIP re: the pytorch 1.7.1 requirement, and tt re: gumbel support.

mazzo23 commented 3 years ago

@nerdyrodent nice, was looking for that answer about cuda10. have to try that before switching os. ...i dont even say it out loud ^^