Batching¶

Utilities for batching and shuffling simulation datasets.

combine_datasets ¶

combine_datasets(datasets)

Concatenate multiple datasets into one.

Parameters:

Name	Type	Description	Default
`datasets`	`Sequence[SimulationDataset]`	List of datasets to concatenate.	required

Returns:

Type	Description
`SimulationDataset`	Combined dataset with entries stacked along the batch dimension.

iterate_minibatches(
    dataset, batch_size, *, drop_last=False
)

Yield mini-batches from a dataset.

Parameters:

Name	Type	Description	Default
`dataset`	`SimulationDataset`	Source dataset.	required
`batch_size`	`int`	Number of samples per batch.	required
`drop_last`	`bool`	Whether to drop the final incomplete batch.	`False`

Yields:

Type	Description
`SimulationDataset`	Mini-batch of the dataset.

shuffle_dataset(dataset, seed=None)

Shuffle dataset entries along the batch dimension.

Parameters:

Name	Type	Description	Default
`dataset`	`SimulationDataset`	Dataset to shuffle.	required
`seed`	`int`	Optional seed for reproducible shuffling.	`None`

Returns:

Type	Description
`SimulationDataset`	Shuffled dataset.