scglue.models.data.Dataset

class scglue.models.data.Dataset(getitem_size=1)[source]

Bases: Dataset

Abstract dataset interface extending that of torch.utils.data.Dataset

Parameters:

getitem_size (int) – Unitary fetch size for each __getitem__ call

Methods

accept_shuffle

Accept shuffling result

clean

Clean up multi-process resources used in custom shuffling

prepare_shuffle

Prepare dataset for custom shuffling

propose_shuffle

Propose shuffling using a given random seed

shuffle

Custom shuffling

shuffle_worker

Background shuffle worker

Attributes

has_workers

Whether background shuffling workers have been registered

logger