streaming.base.partition#

Apportion shards/samples to nodes/ranks/workers for elastically deterministic sample order.

Functions

`get_partitions`	Partition the given number of samples to nodes, ranks, and workers.
`get_partitions_orig`	Partition the given number of samples to nodes, ranks, and workers.
`get_partitions_relaxed`	Partition the given number of samples to nodes, ranks, and workers.

Classes

alias of numpy.ndarray[Any, numpy.dtype[numpy._typing._array_like._ScalarType_co]]