# tf.keras.utils.experimental.DatasetCreator

<br />

|---|
| [![](https://www.tensorflow.org/images/GitHub-Mark-32px.png) View source on GitHub](https://github.com/keras-team/keras/tree/v2.13.1/keras/utils/dataset_creator.py#L24-L116) |

Object that returns a [`tf.data.Dataset`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/data/Dataset) upon invoking.

    tf.keras.utils.experimental.DatasetCreator(
        dataset_fn, input_options=None
    )

[`tf.keras.utils.experimental.DatasetCreator`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/keras/utils/experimental/DatasetCreator) is designated as a supported
type for `x`, or the input, in [`tf.keras.Model.fit`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/keras/Model#fit). Pass an instance of
this class to `fit` when using a callable (with a `input_context` argument)
that returns a [`tf.data.Dataset`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/data/Dataset).

    model = tf.keras.Sequential([tf.keras.layers.Dense(10)])
    model.compile(tf.keras.optimizers.SGD(), loss="mse")

    def dataset_fn(input_context):
      global_batch_size = 64
      batch_size = input_context.get_per_replica_batch_size(global_batch_size)
      dataset = tf.data.Dataset.from_tensors(([1.], [1.])).repeat()
      dataset = dataset.shard(
          input_context.num_input_pipelines, input_context.input_pipeline_id)
      dataset = dataset.batch(batch_size)
      dataset = dataset.prefetch(2)
      return dataset

    input_options = tf.distribute.InputOptions(
        experimental_fetch_to_device=True,
        experimental_per_replica_buffer_size=2)
    model.fit(tf.keras.utils.experimental.DatasetCreator(
        dataset_fn, input_options=input_options), epochs=10, steps_per_epoch=10)

[`Model.fit`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/keras/Model#fit) usage with `DatasetCreator` is intended to work across all
[`tf.distribute.Strategy`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/distribute/Strategy)s, as long as [`Strategy.scope`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/distribute/MirroredStrategy#scope) is used at model
creation:

    strategy = tf.distribute.experimental.ParameterServerStrategy(
        cluster_resolver)
    with strategy.scope():
      model = tf.keras.Sequential([tf.keras.layers.Dense(10)])
    model.compile(tf.keras.optimizers.SGD(), loss="mse")

    def dataset_fn(input_context):
      ...

    input_options = ...
    model.fit(tf.keras.utils.experimental.DatasetCreator(
        dataset_fn, input_options=input_options), epochs=10, steps_per_epoch=10)

> [!NOTE]
> **Note:** When using `DatasetCreator`, `steps_per_epoch` argument in [`Model.fit`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/keras/Model#fit) must be provided as the cardinality of such input cannot be inferred.

<br />

<br />

<br />

| ## Args ||
|---|---|
| `dataset_fn` | A callable that takes a single argument of type [`tf.distribute.InputContext`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/distribute/InputContext), which is used for batch size calculation and cross-worker input pipeline sharding (if neither is needed, the `InputContext` parameter can be ignored in the `dataset_fn`), and returns a [`tf.data.Dataset`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/data/Dataset). |
| `input_options` | Optional [`tf.distribute.InputOptions`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/distribute/InputOptions), used for specific options when used with distribution, for example, whether to prefetch dataset elements to accelerator device memory or host device memory, and prefetch buffer size in the replica device memory. No effect if not used with distributed training. See [`tf.distribute.InputOptions`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/distribute/InputOptions) for more information. |

<br />

## Methods

### `__call__`

[View source](https://github.com/keras-team/keras/tree/v2.13.1/keras/utils/dataset_creator.py#L107-L116)

    __call__(
        *args, **kwargs
    )

Call self as a function.