Simultaneous Trainer

• Source: /trainers/simultaneous_trainer.py

d_grads,g_grads = self.calculate_gradients(D, G)
self.train_d(d_grads)
self.train_g(g_grads)

examples

• Configurations: /trainers/simultaneous_trainer/

{
  "class": "function:hypergan.trainers.simultaneous_trainer.SimultaneousTrainer",
  "optimizer": {
    "class": "function:torch.optim.Adam",
    "lr": 1e-4,
    "betas":[0.0,0.999]
  },
  "hooks": [
    {
      "class": "function:hypergan.train_hooks.adversarial_norm_train_hook.AdversarialNormTrainHook",
      "gamma": 100,
      "loss": ["d"]
    },
    {
      "class": "function:hypergan.train_hooks.negative_momentum_train_hook.NegativeMomentumTrainHook",
      "gamma": 0.33
    }
  ]
}

options

attribute	description	type
optimizer	Optimizer configuration	Config (required)
hooks	Train Hooks	Array of configs (optional)