Open codes for KNDA-GAN would coming soon

python train.py --dataroot yourDataroot --name nameThisTraining
--batch_size 4 --dataset_mode approximateAligned --direction AtoB
--gan_mode lsgan --gpu_ids 0
--lambda_A 10.0 --lambda_B 10.0 --lambda_identity 0.5
--lambda_L1 100.0 --lambda_L2 3.0 --lambda_LG 20.0
--model KNDA --n_epochs 50 --n_epochs_decay 50
--norm instance --preprocess None
--input_nc 1 --output_nc 1
--display_port 8089 --display_env training --pool_size 0
--no_flip

python test.py --dataroot yourDataroot --name yourTrainingModelSaveName
--model YourModel --input_nc 1 --output_nc 1 --preprocess None --batch_size 1
--gpu_ids 1 --phase test --direction AtoB --serial_batches
--results_dir yourResultRoot
--epoch 50 --num_test 1000 """ Here input_nc and output_nc denote the input and output channel of generators epoch choose a epoch to test num_test max test number """

Our code is inspired by CycleGAN2017&Pix2Pix\ PGAN-CGAN