tfds.builder_from_directory | TensorFlow Datasets (original) (raw)
tfds.builder_from_directory
Loads a tfds.core.DatasetBuilder from the given generated dataset path.
tfds.builder_from_directory(
builder_dir: tfds.typing.PathLike
) -> tfds.core.DatasetBuilder
Reconstructs the tfds.core.DatasetBuilder without requiring the original generation code.
From <builder_dir>/features.json it infers the structure (feature names, nested dict,...) and content (image, sequence,...) of the dataset. The serialization format is defined intfds.features.FeatureConnector in to_json().
| Args | |
|---|---|
| builder_dir | str, path of the directory containing the dataset to read ( e.g. ~/tensorflow_datasets/mnist/3.0.0/). |
| Returns | |
|---|---|
| builder | tfds.core.DatasetBuilder, builder for dataset at the given path. |
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2024-04-26 UTC.