Help protect the Great Barrier Reef with TensorFlow on Kaggle Join Challenge


TensorFlow 1 version View source on GitHub

Synchronous training on TPUs and TPU Pods.

Inherits From: Strategy

To construct a TPUStrategy object, you need to run the initialization code as below:

resolver = tf.distribute.cluster_resolver.TPUClusterResolver(tpu='')
strategy = tf.distribute.experimental.TPUStrategy(resolver)

While using distribution strategies, the variables created within the strategy's scope will be replicated across all the replicas and can be kept in sync using all-reduce algorithms.

To run TF2 programs on TPUs, you can either use .compile and .fit APIs in tf.keras with TPUStrategy, or write your own customized training loop by calling directly. Note that TPUStrategy doesn't support pure eager execution, so please make sure the function passed into is a tf.function or is called inside a tf.function if eager behavior is enabled.

tpu_cluster_resolver A tf.distribute.cluster_resolver.TPUClusterResolver, which provides information about the TPU cluster.
device_assignment Optional tf.tpu.experimental.DeviceAssignment to specify the placement of replicas on the TPU cluster.

cluster_resolver Returns the cluster resolver associated with this strategy.

tf.distribute.experimental.TPUStrategy provides the associated tf.distribute.cluster_resolver.ClusterResolver. If the user provides one in __init__, that instance is returned; if the user does not, a default tf.distribute.cluster_resolver.TPUClusterResolver is provided.

extended tf.distribute.StrategyExtended with additional methods.
num_replicas_in_sync Returns number of replicas over which gradients are aggregated.



View source

Adds annotation that tensor will be assigned to a logical device.

# Initializing TPU system with 2 logical devices and 4 replicas.
resolver = tf.distribute.cluster_resolver.TPUClusterResolver(tpu='')
topology = tf.tpu.experimental.initialize_tpu_system(resolver)
device_assignment =
    computation_shape=[1, 1, 1, 2],
strategy = tf.distribute.TPUStrategy(
    resolver, experimental_device_assignment=device_assignment)
iterator = iter(inputs)

def step_fn(inputs):
  output = tf.add(inputs, inputs)

  # Add operation will be executed on logical device 0.
  output = strategy.experimental_assign_to_logical_device(output, 0)
  return output, args=(next(iterator),))

tensor Input tensor to annotate.
logical_device_id Id of the logical core to which the tensor will be assigned.

ValueError The logical device id presented is not consistent with total number of partitions specified by the device assignment.

Annotated tensor with idential value as tensor.


View source

Creates tf.distribute.DistributedDataset from

The returned tf.distribute.DistributedDataset can be iterated over similar to how regular datasets can. NOTE: The user cannot add any more transformations to a tf.distribute.DistributedDataset.

The following is an example:

strategy = tf.distribute.MirroredStrategy()

# Create a dataset
dataset = dataset_ops.Dataset.TFRecordDataset([
  "/a/1.tfr", "/a/2.tfr", "/a/3.tfr", "/a/4.tfr"])

# Distribute that dataset
dist_dataset = strategy.experimental_distribute_dataset(dataset)

# Iterate over the `tf.distribute.DistributedDataset`
for x in dist_dataset:
  # process dataset elements, args=(x,))

In the code snippet above, the tf.distribute.DistributedDataset dist_dataset is batched by GLOBAL_BATCH_SIZE, and we iterate through it using for x in dist_dataset. x a tf.distribute.DistributedValues containing data for all replicas, which aggregates to a batch of GLOBAL_BATCH_SIZE. will take care of feeding the right per-replica data in x to the right replica_fn executed on each replica.

What's under the hood of this method, when we say the instance - dataset - gets distributed? It depends on how you set the through By default, it is set to