Can d_loss and g_loss both become small?

[blgimagineb]

During the training of acgan, I encountered a confusing issue.
After training for more than 10,000 epoch, both the d_loss and g_loss became very small. Additionally, when I added metrics to observe the combined training, I found that the accuracy of both the discriminator and the combined model reached 100%. This seems contradictory to me.