|---| | [ View source on GitHub](https://github.com/tensorflow/transform/blob/v1.21.0/tensorflow_transform/annotators.py#L230-L246) | Creates mapping between user-defined keys and SavedModel assets. tft.annotate_asset( asset_key: str, asset_filename: str ) This mapping is made available in `BeamDatasetMetadata` and is also used to resolve vocabularies in [`tft.TFTransformOutput`](https://www.tensorflow.org/tfx/transform/api_docs/python/tft/TFTransformOutput). > [!NOTE] > **Note:** multiple mappings for the same key will overwrite the previous one.


| ## Args || |---|---|
*** ** * ** *** asset_key: The key to associate with the asset. asset_filename: The filename as it appears within the assets/ subdirectory. Must be sanitized and complete (e.g. include the tfrecord.gz for suffix appropriate files).