Closed razorx89 closed 1 year ago
We had initially written the code that is only compatible with one-to-one synthesis tasks. Since we only dealt with such tasks in the paper, that was the case. We are planning to release v2.0 of our code soon, which will include training/testing with multi-channel input/outputs.
Hey, thanks for this contribution to the scientific community. Could you please add multi-channel support (different for both domains)? I think it should be quite easy to add (define_netG/define_netD have parameters for this), but I haven't had a look at all parts of the code. The dataset loader should also support a 4th dimension for the channels.
Thanks in advance.