tf.raw_ops.ParallelInterleaveDatasetV4

Creates a dataset that applies f to the outputs of input_dataset.

tf.raw_ops.ParallelInterleaveDatasetV4(
    input_dataset,
    other_arguments,
    cycle_length,
    block_length,
    buffer_output_elements,
    prefetch_input_elements,
    num_parallel_calls,
    f,
    output_types,
    output_shapes,
    deterministic='default',
    metadata='',
    name=None
)

The resulting dataset is similar to the InterleaveDataset, except that the dataset will fetch records from the interleaved datasets in parallel.

The tf.data Python API creates instances of this op from Dataset.interleave() when the num_parallel_calls parameter of that method is set to any value other than None.

By default, the output of this dataset will be deterministic, which may result in the dataset blocking if the next data item to be returned isn't available. In order to avoid head-of-line blocking, one can either set the deterministic attribute to "false", or leave it as "default" and set the experimental_deterministic parameter of tf.data.Options to False. This can improve performance at the expense of non-determinism.

input_dataset A Tensor of type variant. Dataset that produces a stream of arguments for the function f.
other_arguments A list of Tensor objects. Additional arguments to pass to f beyond those produced by input_dataset. Evaluated once when the dataset is instantiated.
cycle_length A Tensor of type int64. Number of datasets (each created by applying f to the elements of input_dataset) among which the ParallelInterleaveDatasetV2 will cycle in a round-robin fashion.
block_length A Tensor of type int64. Number of elements at a time to produce from each interleaved invocation of a dataset returned by f.
buffer_output_elements A Tensor of type int64. The number of elements each iterator being interleaved should buffer (similar to the .prefetch() transformation for each interleaved iterator).
prefetch_input_elements A Tensor of type int64. Determines the number of iterators to prefetch, allowing buffers to warm up and data to be pre-fetched without blocking the main thread.
num_parallel_calls A Tensor of type int64. Determines the number of threads that should be used for fetching data from input datasets in parallel. The Python API tf.data.experimental.AUTOTUNE constant can be used to indicate that the level of parallelism should be autotuned.
f A function decorated with @Defun. A function mapping elements of input_dataset, concatenated with other_arguments, to a Dataset variant that contains elements matching output_types and output_shapes.
output_types A list of tf.DTypes that has length >= 1.
output_shapes A list of shapes (each a tf.TensorShape or list of ints) that has length >= 1.
deterministic An optional string. Defaults to "default". A string indicating the op-level determinism to use. Deterministic controls whether the interleave is allowed to return elements out of order if the next element to be returned isn't available, but a later element is. Options are "true", "false", and "default". "default" indicates that determinism should be decided by the experimental_deterministic parameter of tf.data.Options.
metadata An optional string. Defaults to "".
name A name for the operation (optional).

A Tensor of type variant.