blazejdolicki

Thanks for your response. Using more data helped to some extent, but after some more digging I realized the real issue. The CLS dataset has three columns - label, summary...

My package versions differ slightly from those in requirements.txt, maybe sacremoses is related: `fire 0.3.0 ` `sacremoses 0.0.38 ` `sentencepiece 0.1.85` `fastai 1.0.47 `

Thanks for a comprehensive response and all the references. In this case, I will wait for the standalone version. Looking forward to it!

Hi @moritzmeister, I was wondering what's the status of Maggy, did you manage to make it standalone? :)

Thanks for letting us know!

@enzoampil could you maybe provide some example code?

Hey Gabriele, thanks for your elaborate answer, this makes it much more clear now! It seems that [e2wrn.py]( https://github.com/QUVA-Lab/e2cnn/blob/master/examples/e2wrn.py) is an improved version of [e2_wide_resnet.py](https://github.com/QUVA-Lab/e2cnn_experiments/blob/master/experiments/models/e2_wide_resnet.py). Apart from some variable renaming...

I have another question related to the comment above. Based on my understanding it seems to me that the group space used in a FieldType should correspond to it's representation....

Thanks, that makes it more clear. I implemented and trained an equivariant model for N=4 rotations, but when I check the probabilities between rotations by 90 degrees (so perfect rotation...

Hey, Is there any progress on this? :) I stumbled upon the same problem and needed a quick fix, so I dug a bit deeper. It seems more complicated than...