I get this error, but all files are fine and also the config.json about moore-animateanyone HOT 9 CLOSED

@gunjahman
https://pytorch.org/

Reinstall torch 2.0.1 using the instructions.

Make sure to uninstall torch first by running pip uninstall torch torchaudio torchvision

from moore-animateanyone.

Hello @gunjahman , have you install complete files of sd-vae-ft-mse? The issue is likely caused by missing the file path, so I suggest you check the path and its configuration in the config file.

from moore-animateanyone.

Thank you for your answer.

I have checked the path in the config file (configs/prompts/animation.yaml) and it is correct (pretrained_vae_path: "./pretrained_weights/sd-vae-ft-mse")
I have in the root a folder named: .pretrained_weights and in there a folder named: sd-vae-ft-mse and in there there are three files: config.json, diffusion_pytorch_model.bin, diffusion_pytorch_model.safetensors.

But I'm getting that error.

Any ideas?

from moore-animateanyone.

the folder name in the root may be pretrained_weights/ instead of .pretrained_weights. Is it right?

from moore-animateanyone.

Good idea. I tried it and now I get a different error: raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

from moore-animateanyone.

This seems you don't install torch with a cuda version. You can first check whether the cuda is enabled by this:

import torch
print(torch.cuda.is_available())

If output is False, you could reinstall torch.

from moore-animateanyone.

Sorry, I'm a newbie. Where do I write:
import torch
print(torch.cuda.is_available())

from moore-animateanyone.

you can create a new .py file within the folder and run it

from moore-animateanyone.

Thank you.
I did it, and you were right, I get FALSE

python check.py
False

What should I do?

from moore-animateanyone.