Least Squares Loss

• From https://arxiv.org/abs/1611.04076

• Source: /losses/least_squares_loss.py

a,b,c = (config.labels or [-1,1,1])
d_loss = 0.5*((d_real - b)**2) + 0.5*((d_fake - a)**2)
g_loss = 0.5*((d_s

examples

• Configurations: /losses/least_squares_loss/

{
  "class": "function:hypergan.losses.least_squares_loss.LeastSquaresLoss"
}

options

attribute	description	type
labels	[a,b,c]. Defaults to [-1,1,1]	array of floats (optional)