torchtnt.utils.data.MultiIterator

class torchtnt.utils.data.MultiIterator(individual_dataloaders: Mapping[str, Union[DataLoader, Iterable]], iteration_strategy: DataIterationStrategy)

MultiIterator defines the iteration logic to get a batch, given batches from all individual dataloaders. iteration_strategy can include accompanying parameters for a particular iterator, like cycling order for the dataloaders.

Parameters:

• individual_dataloaders (Mapping[str, Union[DataLoader, Iterable]]) – A mapping of DataLoaders or Iterables with dataloader name as key and dataloader/iterable object as value.
• iteration_strategy (DataIterationStrategy) – A dataclass indicating how the dataloaders are iterated over.

Note

TorchData (https://pytorch.org/data/beta/index.html) also has generic multi-data

sources reading support to achieve the same functionality provided by MultiIterator.

For example, mux, mux_longest, cycle, zip etc. Please refer to the documentation for more details.

__init__(individual_dataloaders: Mapping[str, Union[DataLoader, Iterable]], iteration_strategy: DataIterationStrategy) → None

Methods

__init__(individual_dataloaders, ...)