Types & classes — github.com/BlackHC/BatchBALD

CodeHub MCPcopy Create free account

hub / github.com/BlackHC/BatchBALD / types & classes

Types & classes28 in github.com/BlackHC/BatchBALD

⨍ Functions 305◇ Types & classes 28

↓ 5 callersClassDataSource

src/dataset_enum.py:35

↓ 5 callersClassTransformedDataset

Transforms a dataset. Arguments: dataset (Dataset): The whole Dataset transformer (LambdaType): (idx, sample) -> transformed

src/transformed_dataset.py:4

↓ 3 callersClassAcquisitionBatch

src/acquisition_batch.py:6

↓ 3 callersClassRandomFixedLengthSampler

Sometimes, you really want to do more with little data without increasing the number of epochs. This sampler takes a `dataset` and draws `ta

src/random_fixed_length_sampler.py:5

↓ 2 callersClassContextStopwatch

src/context_stopwatch.py:6

↓ 2 callersClassSamplerModel

src/sampler_model.py:73

↓ 1 callersClassActiveLearningData

Splits `dataset` into an active dataset and an available dataset.

src/active_learning_data.py:7

↓ 1 callersClassAggregateAccuracies

src/al_notebook/results_loader.py:31

↓ 1 callersClassExperimentData

src/dataset_enum.py:25

↓ 1 callersClassForwarder

src/test_acquisition_functions.py:87

↓ 1 callersClassIgniteProgressBar

src/ignite_progress_bar.py:6

↓ 1 callersClassLoaders

src/recover_model.py:19

↓ 1 callersClassNoDropoutModel

src/sampler_model.py:61

↓ 1 callersClassRecoveredModel

src/recover_model.py:25

↓ 1 callersClassSubsetEvalResults

src/reduced_consistent_mc_sampler.py:16

↓ 1 callersClassTrainModelResult

src/train_model.py:13

↓ 1 callersClassVGG

VGG with BatchNorm performs best. We only add MCDropout in the classifier head (where VGG used dropout before, too).

src/vgg_model.py:23

↓ 1 callersClassVIPArgs

src/al_notebook/results_loader.py:496

ClassAcquisitionFunction

src/acquisition_functions.py:34

ClassAcquisitionMethod

src/acquisition_method.py:11

ClassBayesianModule

A module that we can sample multiple times from given a single input batch. To be efficient, the module allows for a part of the forward pass to

src/mc_dropout.py:16

ClassBayesianNet

src/emnist_model.py:9

ClassBayesianNet

src/mnist_model.py:8

ClassDatasetEnum

src/dataset_enum.py:89

r"""Randomly zeroes some of the elements of the input tensor with probability :attr:`p` using samples from a Bernoulli distribution. The eleme

src/mc_dropout.py:125

ClassMCDropout2d

r"""Randomly zeroes whole channels of the input tensor. The channels to zero-out are randomized on every forward call. During eval time, a fi

src/mc_dropout.py:161

ClassRestoringScoreGuard

RestoringScoreGuard handler can be used to stop the training if no improvement after a given number of events Args: patience (int):

src/ignite_restoring_score_guard.py:7

Class_MCDropout

src/mc_dropout.py:68