not training

[tryfonaskam]

in google colaboration(free) after going through this tutorial and this guide i start the "Perform Initial Training" and i get this error /content/stylegan2-ada-pytorch/torch_utils/ops/conv2d_gradfix.py:55: UserWarning: conv2d_gradfix not supported on PyTorch 2.3.1+cu121. Falling back to torch.nn.functional.conv2d(). warnings.warn(f'conv2d_gradfix not supported on PyTorch {torch.__version__}. Falling back to torch.nn.functional.conv2d().')

[Neilstid]

You should retry with an older version of pytorch and see if it still appears. The recommended version of pytorch for this repo is pytorch=1.9.1. I personally use pytorch=1.13.1 and it works fine too. Downgrading to pytorch 1.xx.x should fix the problem. Hope it will help you :)

[tryfonaskam]

I am going to try this and share my results

[tryfonaskam]

i did downgrade my pytorch version to 1.13.1 and now i am geting this error AttributeError: module 'scipy.linalg' has no attribute 'tril' i also want to say this is not the only error i get this 2024-08-05 11:20:30.541072: W external/org_tensorflow/tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/lib64-nvidia and this /content/stylegan2-ada-pytorch/torch_utils/ops/conv2d_gradfix.py:55: UserWarning: conv2d_gradfix not supported on PyTorch 2.3.1+cu121. Falling back to torch.nn.functional.conv2d(). warnings.warn(f'conv2d_gradfix not supported on PyTorch {torch.__version__}. Falling back to torch.nn.functional.conv2d().')

[Neilstid]

First the scipy error can be fixed by downgrading it to <=1.12.0 (https://github.com/octo-models/octo/issues/71)

The second error is related to a cuda error. More details here:

• how-to-fix-tensorflow-warning-could-not-load-dynamic-library-cudart64
• Could not load dynamic library 'libcudart.so.11.0';

If you have not done it yet, you should get cudatoolkit to solve this (conda install cudatoolkit). However, if you already have, look at the ressource above, they might help.

I'm quite surprise about the third error telling the pytorch version is 2.3.1 :/ may be cached data? But I think if you run: import torch print(torch.__version__)

this will likely return you 2.3.1+cu121 :/

[tryfonaskam]

the pytorch version is not staying the same for some reason I will take a look at the links you provided to me and maybe do some troubleshooting myself if I have any updates I will post them here

[tryfonaskam]

I tried a lot of stuff changing settings but nothing is working.If anyone finds a solution pls comment on this