maxibove13/wakegan

Application of a cDCGAN to predict the flow around a WT conditioned to the inflow velocity

wakegan's Issues

create loss class to make plots in a cleaner way

solve minibatch training issue

when training with a single batch, training and testing data achieve same rmse, when testing should be much worse, and the difference between them worse than when training with all the batches. There must be an issue with the implementation.

enable 2 channel training and testing

monitor error in m/s

create training gif of a single testing example