GDUMB LIDER#

Arguments#

Options

--alpha_lip_lambdafloat

Help: Lambda parameter for lipschitz minimization loss on buffer samples

  • Default: 0

--beta_lip_lambdafloat

Help: Lambda parameter for lipschitz budget distribution loss

  • Default: 0

--headless_init_actstr

Help: None

  • Default: relu

  • Choices: relu, lrelu

--grad_iter_stepint

Help: Step from which to enable gradient computation.

  • Default: -2

--maxlrfloat

Help: Max learning rate.

  • Default: 0.05

--minlrfloat

Help: Min learning rate.

  • Default: 0.0005

--fitting_epochsint

Help: Number of epochs to fit the buffer.

  • Default: 256

--cutmix_alphafloat

Help: Alpha parameter for cutmix

  • Default: 1.0

Rehearsal arguments

Arguments shared by all rehearsal-based methods.

--buffer_sizeint

Help: The size of the memory buffer.

  • Default: None

--minibatch_sizeint

Help: The batch size of the memory buffer.

  • Default: None

Classes#

class models.gdumb_lider.GDumbLider(backbone, loss, args, transform, dataset=None)[source]#

Bases: LiderOptimizer

GDumb learns an empty model only on the buffer. Treated with LiDER!

COMPATIBILITY: List[str] = ['class-il', 'task-il']#
NAME: str = 'gdumb_lider'#
end_task(dataset)[source]#
static get_parser(parser)[source]#
Return type:

ArgumentParser

observe(inputs, labels, not_aug_inputs, epoch=None)[source]#

Functions#

models.gdumb_lider.fit_buffer(self, epochs)[source]#