Logistic Loss

• Adapted from http://stylegan.xyz/paper

• Source: /losses/logistic_loss.py

d_loss = self.softplus(-d_real) + self.softplus(d_fake)
g_loss = self.softplus(-d_fake)

examples

• Configurations: /losses/logistic_loss/

{                                                                                       
  "class": "function:hypergan.losses.logistic_loss.LogisticLoss"
}

options

attribute	description	type
beta	https://pytorch.org/docs/stable/_modules/torch/nn/modules/activation.html#Softplus	float (optional)
threshold	https://pytorch.org/docs/stable/_modules/torch/nn/modules/activation.html#Softplus	float (optional)