imaginaire.optimizers package

Submodules

imaginaire.optimizers.fromage module

class imaginaire.optimizers.fromage.Fromage(params, lr=<required parameter>, momentum=0)

Bases: torch.optim.optimizer.Optimizer

Fromage optimizer implementation (https://arxiv.org/abs/2002.03432)

step(closure=None)

Performs a single optimization step.

Parameters

closure (callable, optional) – A closure that reevaluates the model and returns the loss.

imaginaire.optimizers.madam module

class imaginaire.optimizers.madam.Madam(params, lr=<required parameter>, scale=3.0, g_bound=None, momentum=0)

Bases: torch.optim.optimizer.Optimizer

MADAM optimizer implementation (https://arxiv.org/abs/2006.14560)

step(closure=None)

Performs a single optimization step.

Parameters

closure (callable, optional) – A closure that reevaluates the model and returns the loss.

Module contents

class imaginaire.optimizers.Fromage(params, lr=<required parameter>, momentum=0)

Bases: torch.optim.optimizer.Optimizer

Fromage optimizer implementation (https://arxiv.org/abs/2002.03432)

step(closure=None)

Performs a single optimization step.

Parameters

closure (callable, optional) – A closure that reevaluates the model and returns the loss.

class imaginaire.optimizers.Madam(params, lr=<required parameter>, scale=3.0, g_bound=None, momentum=0)

Bases: torch.optim.optimizer.Optimizer

MADAM optimizer implementation (https://arxiv.org/abs/2006.14560)

step(closure=None)

Performs a single optimization step.

Parameters

closure (callable, optional) – A closure that reevaluates the model and returns the loss.