Problems of the training curve

AlexHex7 / SimGAN_pytorch

[Refer to wayaai/SimGAN(Keras&Tensorflow)] Implementation of Apple's Learning from Simulated and Unsupervised Images through Adversarial Training

41 stars 9 forks source link

Problems of the training curve #2

Open automan000 opened 6 years ago

automan000 commented 6 years ago

Hi, AlexHex7. I forked your code and did several changes for bug fixing and speed-up. But my training procedure wasn't pleasant. I tested many hyper-parameters but the results were still terrible. Could you share your training log with me? Thanks a lot.

AlexHex7 commented 6 years ago

Hi~ In fact, last year when I did this experiments, it was also unstable. I have not followed the research area of GAN for a long time. I think you can try yo introduce training techniques from other GAN paper to do this experiments, such as more stable loss function, VAE+GAN ...