Skip to main content

dataloaders

PyTorch-specific DataLoader implementations.

Classes​

PyTorchIterableBitfountDataLoader​

class PyTorchIterableBitfountDataLoader(    dataset: _IterableBitfountDataset,    secure_rng: bool = False,    batch_size: int = 1,    shuffle: bool = False,):

Wraps a PyTorch DataLoader with bitfount functions.

Arguments

  • batch_size: The batch size for the dataloader. Defaults to 1.
  • dataset: An iterable dataset.
  • secure_rng: A boolean value indicating whether to use a secure random number generator. Defaults to False.
  • shuffle: A boolean value indicating whether the values in the dataset should be shuffled. Defaults to False.

Attributes

  • batch_size: The batch size for the dataloader. Defaults to 1.
  • secure_rng: A boolean value indicating whether to use a secure random number generator. Defaults to False.
  • shuffle: A boolean value indicating whether the values in the dataset should be shuffled. Defaults to False.

Ancestors​

Variables​

  • nominal_num_batches : int - Batch count ignoring frame-shape bucketing.

    Epoch arithmetic only - not the number of batches iteration will actually yield for image datasets, where batches are grouped by frame shape and can be smaller than batch_size.

    Returns: The number of elements when there is no batch size, otherwise len(dataset) divided by the batch size, rounded up.

Methods​

expect_key_in_iter​

def expect_key_in_iter(self) ‑> bool:

Will there be a data key entry in the output from iteration?