An op that demultiplexes a tensor to be sharded by XLA to a list of partitioned
View aliases
Compat aliases for migration
See
Migration guide for
more details.
`tf.compat.v1.tpu_partitioned_output_v2`
tf.tpu_partitioned_output_v2(
inputs, num_splits, partition_dims, name=None
)
outputs outside the XLA computation. Supports ND sharding.
Args |
inputs
|
A Tensor .
A tensor which represents the full shape of partitioned tensors.
|
num_splits
|
An int that is >= 1 .
|
partition_dims
|
A list of ints .
A list of integers describing how each dimension is partitioned. Emptiness
indicates the inputs are replicated.
|
name
|
A name for the operation (optional).
|
Returns |
A list of num_splits Tensor objects with the same type as inputs .
|