BIC#

Arguments#

Options

--bic_epochsint

Help: bias injector.

  • Default: 250

--tempfloat

Help: softmax temperature

  • Default: 2.0

--valset_splitfloat

Help: bias injector.

  • Default: 0.1

--wd_regfloat

Help: bias injector.

  • Default: None

--distill_after_bic0|1|True|False -> bool

Help: None

  • Default: 1

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.bic.BiC(backbone, loss, args, transform, dataset=None)[source]#

Bases: ContinualModel

Bias Correction.

COMPATIBILITY: List[str] = ['class-il', 'task-il']#
NAME: str = 'bic'#
begin_task(dataset)[source]#
build_buffer(dataset)[source]#
end_task(dataset)[source]#
evaluate_bias(fprefx)[source]#
forward(x, anticipate=False)[source]#
static get_parser(parser)[source]#
Return type:

ArgumentParser

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