Open ShacharWeis opened 1 year ago
I think that is because you run computation on CPU instead of GPU. Can you print the value of thedevice
variable?
Yes, that was the problem. I installed torch with CUDA and now it works.
Here is the fix:
pip installing torch pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
Yes, that was the problem. I installed torch with CUDA and now it works.
Here is the fix:
pip installing torch pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
For me it still doesn't work, I have an RTX 3060 12G and Ryzen 5 2600, isn't it because of the python version? I'm using Python 3.11.3, let me guys know what version are you using
@BryanValc Can you please write what is value of device
variable?
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print(device)
@Kabanosk I did that, but it keeps using cpu
You have to uninstall torch and then install a CUDA version of torch. Then the variable value is "cuda"
Or, you can just not bother, because the 3D models output from this AI are complete garbage. Really. The examples they show here were carefully hand picked. It's a waste of time (in it's current state). I played with it for an hour and gave up. Nothing it makes is even remotely unusable.
You have to uninstall torch and then install a CUDA version of torch. Then the variable value is "cuda"
Or, you can just not bother, because the 3D models output from this AI are complete garbage. Really. The examples they show here were carefully hand picked. It's a waste of time (in it's current state). I played with it for an hour and gave up. Nothing it makes is even remotely unusable.
I tried uninstalling it, then running this command: pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
But still nothing.. maybe you're right, it isn't much of a big deal as of now, it is like the beginning of stable diffusion
You may need to delete your vevn folder or create a new venv. If the requirement is already satisfied according to pip, it won't install the cuda version.
pip uninstall torch
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
Then restart your kernel
I cloned the repo and installed the packages needed. Running on a AMD Ryzen 9 laptop with an RTX3070 at 16gb RAM
I run a text-to-model sample and I see this: 0%| | 0/64 [00:00<?, ?it/s]
after 15 minutes it's still at 0% Is it stuck or just really slow?