Mikhail Merkulov

@Runescaped I've tried the same code on not virtual Ubuntu and it works. Since my primary OS is Windows, I just thought someone already faced and fixed the same problem

for every images I've tried.

Cool, thanks! I'll try it in a few days when I get back to Windows)

I'm sure that I haven't used custom layers at that time, so I think I'll put some debugging messages to see if this happens on Windows. Thanks for help!

@htoyryla looks like you're right and there are wrong layers. The difference between my code and original code is that I've fixed model load for Windows: ` local ok, checkpoint...

I think the problem is that according to the [documentation](https://github.com/torch/torch7/blob/master/doc/serialization.md) Torch by default saves models in binary format, which is platform dependent. I will try to convert the model file...

CUDA driver version is insufficient for CUDA runtime version. Check your CUDA driver and runtime version. [This](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/#axzz4Sx6HmFRU) guide might be helpful.

My experiments (not vary familiar with ML) showed some options for this problem - Try to play with style_scale and style_layers. I have python scripts that try almost all valuable...

My last time I've used these options, but the result depends on the styling image anyway: `python run_neural_style.py -style_scale 1.2 -init image -style_image .... -content_image ~... -image_size 750 -output_path ......

@danielgerlag I'd like to try to contribute, but I lack understanding of some details. How do I save some error counter information in the step? As far as I understand...