datasets¶

BatchesMixin¶

class BatchesMixin¶

mixin for supporting batches and collate method from datasets.

batches(batch_size=32, shuffle=False, drop_last=False, num_workers=0, **kwargs)¶

return a Dataloader that iterates over batches of this dataset.

Note

this method supports additional keyword args which will be passed to the dataloader.

Parameters

batch_size (int) – Defaults to 32.
shuffle (bool) – if True shuffle the dataset. Defaults to False.
drop_last (bool) – if true drop the last batch if it is less than batch size. Defaults to False.
num_workers (int) – number of workers. Defaults to 0.

Return type

DataLoader

class TensorDataset(*tensors)¶

class XYDataset(x, y)¶

basic dataset that returns a tuple of inputs and targets.

supports hearth.containers.TensorDataset