RuntimeError: solve: MAGMA library not found in compilation. Please rebuild with MAGMA

Has anyone ever run into this error? I see on PyTorch boards that it has to do with Conda installs but that’s not something I have control over since I’m running it on Kaggle.

Here is my notebook. If anyone has any idea I’d be super thankful for the help!

i gave your notebook a shot and saw the same errors, even after installing MAGMA with conda.

!conda install -y -c pytorch magma-cuda110

maybe try training in a colab or an env where you have more control.

I have to be able to run it on Kaggle to compete. I wish that was an option since it runs fine in Colab.

Getting the same error when using GPU in kaggle. Not getting it if I use CPU compute in kaggle kernels.

If anyone runs into this. The solution is here:

You have to downgrade to pytorch 1.9.0 until there is a fix.

This issue is still there.
I am trying to replicate the bear classification problem from book but I am getting the same problem when I am running the following code on GPU:

bears_data_block = = Resize(128), batch_tfms = aug_transforms(mult=2))
dls = bears_data_block.dataloaders(path)
dls.train.show_batch(max_n=4, nrows=1)

I tried almost following things as suggested in the above mentioned Kaggle discussion thread but nothing worked out.
!conda install -c pytorch magma-cuda110 -y
pip install --user torch==1.9.0 torchvision==0.10.0 torchaudio==0.9.0 torchtext==0.10.0
pip install --user torch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 torchtext==0.8.1.

Can anyone please suggest how to solve it?