EnsembleIndexedDataset

class bayesflow.datasets.EnsembleIndexedDataset(dataset: PyDataset, member_names: Sequence[str], data_reuse: float = 1.0, drop_last: bool | None = None, **kwargs)

Bases: PyDataset

Ensemble batches drawn from member-specific windows into an indexable dataset.

See EnsembleDataset, which is the recommended entry point.

Parameters:

datasetkeras.utils.PyDataset

An indexable BayesFlow dataset (OfflineDataset, DiskDataset).

member_namesSequence[str]

Names of ensemble members, used as dictionary keys.

data_reusefloat, default=1.0

Degree of independence between ensemble members in [0, 1].

drop_lastbool, optional

Whether to drop the last step of each epoch if the member windows have fewer than batch_size samples left. If None (the default), the setting is inherited from the wrapped dataset.

on_epoch_end()

Method called at the end of every epoch.

property max_queue_size

property num_batches

Number of batches in the PyDataset.

Returns:

The number of batches in the PyDataset or None to indicate that the dataset is infinite.

on_epoch_begin()

Method called at the beginning of every epoch.

property use_multiprocessing

property workers