Skip to main content

databunch

Classes concerning databunches.

Classes​

BitfountDataBunch​

class BitfountDataBunch(    data_structure: DataStructure,    schema: BitfountSchema,    datasource: BaseSource,    data_splitter: Optional[DatasetSplitter] = None,    data_factory: Optional[_DataFactory] = None,    ignore_support_cols: bool = False,):

Wrapper for train, validation and test dataloaders.

Provides methods to access dataloaders for training and evaluation. This is strictly a model concept and is not necessary for algorithms that do not have models.

Arguments

  • data_structure: A DataStructure object.
  • schema: A TableSchema object.
  • datasource: A BaseSource object.
  • data_factory: A _DataFactory instance for creating datasets and dataloaders. Defaults to None.

Methods​

get_test_dataloader​

def get_test_dataloader(    self, batch_size: Optional[int] = None, **kwargs: Any,) ‑> Optional[BitfountDataLoader]:

Gets the relevant data loader for test data.

get_train_dataloader​

def get_train_dataloader(    self, batch_size: Optional[int] = None, **kwargs: Any,) ‑> Optional[BitfountDataLoader]:

Gets the relevant data loader for training data.

get_validation_dataloader​

def get_validation_dataloader(    self, batch_size: Optional[int] = None, **kwargs: Any,) ‑> Optional[BitfountDataLoader]:

Gets the relevant data loader for validation data.