# Module: tf.compat.v1.tpu

<br />

Ops related to Tensor Processing Units.

## Modules

[`experimental`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/experimental) module: Public API for tf.tpu.experimental namespace.

## Classes

[`class CrossShardOptimizer`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/CrossShardOptimizer): An optimizer that averages gradients across TPU shards.

[`class PaddingSpec`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/PaddingSpec): Represents the type of padding policies for tpu.replicate.

[`class XLAOptions`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/tpu/XLAOptions): XLA compilation options.

## Functions

[`batch_parallel(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/batch_parallel): Shards `computation` along the batch dimension for parallel execution.

[`bfloat16_scope(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/bfloat16_scope): Scope class for bfloat16 variables so that the model uses custom getter.

[`core(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/core): Returns the device name for a core in a replicated TPU computation.

[`cross_replica_sum(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/cross_replica_sum): Sum the input tensor across replicas according to group_assignment.

[`initialize_system(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/initialize_system): Initializes a distributed TPU system for use with TensorFlow.

[`outside_compilation(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/outside_compilation): Builds part of a computation outside any current TPU replicate scope.

[`replicate(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/replicate): Builds a graph operator that runs a replicated TPU computation.

[`rewrite(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/rewrite): Rewrites `computation` for execution on a TPU system.

[`shard(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/shard): Shards `computation` for parallel execution.

[`shutdown_system(...)`](https://www.tensorflow.org/versions/r2.13/api_docs/python/tf/compat/v1/tpu/shutdown_system): Shuts down a running a distributed TPU system.