How can I use a loss as the stopping criteria in Wasserstein GAN?

[RamziFsm]

Throughout the training step, the variation of generator loss and critic loss for 1000 epochs as following:

Does this variation look like correct?
How can I use a loss as the stopping criteria in Wasserstein GAN? Can I use Generator loss or critic loss? Can I use the early stopping method?
Thanks