load_dataset(train: bool = True, batch_size: Optional[int] = None, shuffle: bool = False, pin_memory: bool = True, object_type: Literal['VisionData', 'DataLoader'] = 'DataLoader', use_iterable_dataset: bool = False, n_samples=None, device: Union[str, device] = 'cpu') Union[DataLoader, VisionData][source]#

Download MNIST dataset.

trainbool, defaultTrue

Train or Test dataset

batch_size: int, optional

how many samples per batch to load

shufflebool , defaultFalse

to reshuffled data at every epoch or not, cannot work with use_iterable_dataset=True

pin_memorybool, defaultTrue

If True, the data loader will copy Tensors into CUDA pinned memory before returning them.

object_typeLiteral[Dataset, DataLoader], default ‘DataLoader’

object type to return. if ‘VisionData’ then deepchecks.vision.VisionData will be returned, if ‘DataLoader’ then torch.utils.data.DataLoader

use_iterable_datasetbool, default False

if True, will use IterableTorchMnistDataset instead of TorchMnistDataset

n_samplesint, optional

Only relevant for loading a VisionData. Number of samples to load. Return the first n_samples if shuffle is False otherwise selects n_samples at random. If None, returns all samples.

devicet.Union[str, torch.device], default‘cpu’

device to use in tensor calculations

Union[:obj:`deepchecks.vision.VisionData`, :obj:`torch.utils.data.DataLoader`]

depending on the object_type parameter value, instance of deepchecks.vision.VisionData or torch.utils.data.DataLoader will be returned