Custom types#

Flytekit ships with an extensible type system to make it easy for anyone to extend and add new types.

Note

To contribute a Flyte type, see the extensibility contribution guide.

StructuredDataset type#

`StructuredDataset`	This is the user facing StructuredDataset class.
`StructuredDatasetEncoder`
`StructuredDatasetDecoder`

`FlyteFile`
`HDF5EncodedFile`	This can be used to denote that the returned file is of type hdf5 and can be received by other tasks that accept an hdf5 format.
`HTMLPage`	Can be used to receive or return an HTMLPage.
`JoblibSerializedFile`	This File represents a file that was serialized using joblib.dump method can be loaded back using joblib.load.
`JPEGImageFile`	Can be used to receive or return an JPEGImage.
`PDFFile`	Can be used to receive or return an PDFFile.
`PNGImageFile`	Can be used to receive or return an PNGImage.
`PythonPickledFile`	This type can be used when a serialized Python pickled object is returned and shared between tasks.
`PythonNotebook`	This type is used to identify a Python notebook file.
`SVGImageFile`	Can be used to receive or return an SVGImage.

`FlyteDirectory`
`TensorboardLogs`	This type can be used to denote that the output is a folder that contains logs that can be loaded in TensorBoard.
`TFRecordsDirectory`	This type can be used to denote that the output is a folder that contains tensorflow record files.

`FlyteIterator`
`JSON`	alias of `Dict`[`str`, `Any`] \| `List`[`Any`] \| `bool` \| `float` \| `int` \| `str`

`PyTorchCheckpoint`	This class is helpful to save a checkpoint.
`PyTorchCheckpointTransformer`	TypeTransformer that supports serializing and deserializing checkpoint.
`PyTorchModuleTransformer`
`PyTorchTensorTransformer`

`TensorFlowRecordFileTransformer`	TypeTransformer that supports serialising and deserialising to and from TFRecord file.
`TensorFlowRecordsDirTransformer`	TypeTransformer that supports serialising and deserialising to and from TFRecord directory.