pfnet/PaintsChainer

about new style of paints chainer

Closed this issue · 2 comments

I saw the new style 'Satsuki' in the web site. Could you tell me what different in training between the new style and the original one. Is it just using different dataset? Or It is using different network structure?

Actually dataset and network structure is almost same.
The difference is mainly loss function and thats param.

Thanks for reply