BatchLoader

BatchLoader: iterator-based batching over any Dataset.

class src.batch_loader.BatchLoader(dataset: Dataset, batch_size: int, shuffle: bool = False, drop_last: bool = False)[source]

Bases: object

Wraps a dataset and yields batches of data via iteration.

Batches are built from indices only; the actual data is loaded from the dataset (which may itself be lazy) only when each batch is consumed.

Parameters:

dataset – Any Dataset instance (labeled or unlabeled).
batch_size – Number of samples per batch.
shuffle – If True, indices are randomly shuffled before batching at the start of every iteration. If False, data is returned in its original order.
drop_last – If True, the last batch is discarded when it contains fewer than batch_size samples. If False (default), the last batch is kept even if it is smaller.

Raises:

Example:

loader = BatchLoader(dataset, batch_size=32, shuffle=True)
for batch in loader:
    # batch is a list of (data, label) tuples or data items
    ...

property shuffle: bool: Whether indices are shuffled at the start of each iteration.