why use the sigmoid？

Hello man,
in model.py
line 111 `self.D_, self.D_logits_ = self.discriminator(self.G, self.y, reuse=True)`
line 311

d_loss

is it a bug ?
may be it should be
sigmoid_cross_entropy_with_logits(self.D_logits, tf.ones_like(self.D_logits)))
and we need not to calculate sigmoid(h4).

i am a noob . if i am wrong, i hope that you get tell me why ？
thanks.

Start by reading model.py around lines 111 and 311, then trace the discriminator output and d_loss formulation. Compare the sigmoid output with the logits passed to TensorFlow's cross-entropy operation; done means confirming whether this is a bug and documenting the explanation or required change.

Written by the indexing model from the issue text.