I am trying to implement a GAN to generate network traffic .csv dataset (tabular GAN) and my training result continued to show [D loss: nan, acc.: 50%] [G loss: nan]. I figured