F Divergence Loss

From https://arxiv.org/abs/1606.00709​
Source: /losses/f_divergence_loss.py​
Configurations: /losses/f_divergence_loss/​
{
  "class": "class:hypergan.losses.f_divergence_loss.FDivergenceLoss",
  "type": "js",
  "g_loss_type": "js",
  "regularizer": "js"
}
options
attribute
description
type
type
supported types js,js_weighted,gan,reverse_kl,pearson,jeffrey,alpha1,alpha2,squared_hellinger,neyman,total_variation,alpha1.  Defaults to gan
string (optional)
g_loss_type
Defaults to type's value
string (optional)
regularizer
Defaults to none.  Same options as type
string (optional)

ALI Loss

Least Squares Loss

Last updated 8 months ago

attribute	description	type
type	supported types `js`,`js_weighted`,`gan`,`reverse_kl`,`pearson`,`jeffrey`,`alpha1`,`alpha2`,`squared_hellinger`,`neyman`,`total_variation`,`alpha1`. Defaults to `gan`	string (optional)
g_loss_type	Defaults to type's value	string (optional)
regularizer	Defaults to none. Same options as type	string (optional)