I am trying WGAN to generate a starry-night style painting. But the model outputs are only noises.
I saw a topic with similar issue. I followed that and set tfms=None, but still not working.
Can someone help here?
il=GANItemList.from_folder(dest,noise_sz=128)
bs,size=(64,128)
data=(il.split_none()
.label_from_func(noop)
.transform(tfms=[[crop_pad(size=size,row_pct=(0,1),col_pct=(0,1))],[]],size=size, tfm_y=True)
.databunch(bs=bs)
.normalize(stats= [torch.tensor([0.5,0.5,0.5]), torch.tensor([0.5,0.5,0.5])], do_x=False, do_y = True)
)
generator=basic_generator(in_size=128,n_channels=3,noise_sz=128,n_extra_layers=1)
critic=basic_critic(in_size=128,n_channels=3, n_extra_layers=1)
learn=GANLearner.wgan(data, generator, critic, switch_eval=False,
opt_func=partial(optim.Adam,betas=(0.,0.99)),wd=0.)
