Du Ang

@8marta It works not well with non-square mask since no training under this setting.

You could try the same scale standard. It should support arbitrary sizes of inputs during testing phrase but there may be some bugs now.

When I implemented the code, I assumed the input size is 256\*256. It is possible that some operation in the contextual attention layer makes it only compatible with the 256\*256...

It seems that it is caused by some different padding in the contextual attention layer. You can compare the feature map's shape of each layer with Yu's version and find...

@kinfeparty Thanks for your question. Yes, the discriminators make it only compatible with fixed sizes of images since there are fully-connected layers at the last layer. You may remove that...

I have no idea about the problem. But I think you can check the network outputs first since the results above are copy-paste from the network outputs. Refers to [here](https://github.com/DAA233/generative-inpainting-pytorch/blob/master/test_single.py#L105).

It is probably because the contextual attention module here doesn't support arbitrary image size, refers to #14.

Could you post your error message or logs here?

@dragen1860 Thanks for your attention! The synced batch norm is not used in this repo. The original tensorflow implementation of deep fill v1 uses only 1 gpu to train the...

In my view, `core dumped` error may happen when your GPU memory is insufficient during training. Maybe you can monitor the GPU memory usage when you train it again. Since...