All Dataset

class olympus.datasets.dataset.AllDataset(dataset, data_path=None, input_shape=None, target_shape=None, train_size=None, valid_size=None, test_size=None, transforms=None)

Bases: torch.utils.data.dataset.Dataset

Olympus data sets are concatenated data sets that includes train, validation and test sets This allow us to change how each sets are splits and give us greater power to design performance tests.

Read more on how Olympus uses custom splits to evaluate model performance at :ref XYZ

Attributes:

dataset: TorchDataset

Underlying dataset (concatenation of original train and test sets)

collate_fn: Optional[Callable] !! static method !!

merges a list of samples to form a mini-batch of Tensor(s). Used when using batched loading from a map-style dataset.

Methods

categories()	Dataset tags so we can filter what we want depending on the task
transforms()

register_datapipe_as_function
register_function

static categories()

Dataset tags so we can filter what we want depending on the task

classes

Return the mapping between samples index and their class

collate_fn = None

dataset = None

input_shape

Return the size of the samples

target_shape

Return the size of the target

test_size

Size of the test set

train_size

Size of the training set

transforms()

valid_size

Size of the validation set