Roman

Hi! Set USER environment variable to any value, see [this issue](https://github.com/saic-mdal/lama/issues/80). You're trying to train, so you might also need TORCH_HOME environment ([this issue](https://github.com/saic-mdal/lama/issues/98)). I do not have a Windows...

Hi! Sorry for the late reply. Does this kind of artifacts appear always? Or maybe there are specific conditions where they are more noticeable than usual? Basically, you can try...

How big is your dataset? What is your training resolution and testing resolution?

Hm... that's weird In which resolution do you feed the images during training? Am I correct that you're applying Lama-Regular model to images of much higher resolution than during training?...

Thank you! We'll check your images!

@ImmortalSdm is this image from training dataloader or from validation? If this is validation, then such artifacts may apper when the mask is non-binary, e.g. it has smooth transition between...

@eduardathome Sorry for the late reply again! Have you tried training in 256? Already after releasing the codebase we found that with Fourier-based generators training in 256 yields more robust...

When fine-tuning a pretrained model, please keep in mind: 1. Overfitting to a single image with discriminator might break because it is too easy for discriminator to remember the real...

Hi! Sorry for the late reply and thank you for the ideas - they are perfectly reasonable and worth doing! We'll try to find time to implement that

Hi! Do you do `export PYTHONPATH="path/to/lama/folder"` as [stated in the readme](https://github.com/saic-mdal/lama#inference-)?