hi about SFT layer

[shz1314]

I am sorry to trouble you, i want to introduce your SFT layer in my image-to-image translation work.If you have already finished SFT layer in pytorch, please send it to me. Very very thanks. My email is [e-mail]
thanks
haoze

[Xintao]

@shz1314, I have just added raw sft-gan testing codes to pytorch repo. Please see the sft-arch.py file in BasicSR repo.

Note that pre-training is import to sft-gan. In our implementation, we first train a normal SR network (with BN layers) under GAN architecture and then we substitute the BN layers with SFT layers (use the BN parameters as the initialization) [BN layers can also be kept and insert new SFT layers].

We will add the training codes in pytorch later. Also, we will explore more stable and effective training.