mixmo.loaders.abstract_loader.AbstractDataLoader

class mixmo.loaders.abstract_loader.AbstractDataLoader(config_args, dataplace, split_test_val=False, corruptions=False)

Bases: object

General dataloader that defines how loaders are built

__init__(config_args, dataplace, split_test_val=False, corruptions=False)

Initialize self. See help(type(self)) for accurate signature.

Methods

__init__(config_args, dataplace[, …])	Initialize self.
make_corruptions_test_dataset()	Make robustness test dataset à la CIFAR10-C Prototype function (redefined for specific datasets)
make_standard_loader(dataset)	Build a dataloader from a dataset (wrapper on torch.utils)

_init_train_loader()

Build the train loader with the proper sampler and data augmentations

_init_valtest_loader(split_test_val)

Build the test (and possibly val) loader with the proper sampler and data augmentations

make_corruptions_test_dataset()

Make robustness test dataset à la CIFAR10-C Prototype function (redefined for specific datasets)

make_standard_loader(dataset)

Build a dataloader from a dataset (wrapper on torch.utils)