Missed TensorFlow World? Check out the recap. Learn more

tf.contrib.distribute.MirroredStrategy

View source on GitHub

Class MirroredStrategy

Mirrors vars to distribute across multiple devices and machines.

Inherits From: Strategy

*** contrib version ***

This strategy uses one replica per device and sync replication for its multi-GPU version.

When cluster_spec is given by the configure method., it turns into the mulit-worker version that works on multiple workers with in-graph replication. Note: configure will be called by higher-level APIs if running in distributed environment.

There are several important concepts for distributed TensorFlow, e.g. client, job, task, cluster, in-graph replication and synchronous training and they have already been defined in the TensorFlow's documentation. The distribution strategy inherits these concepts as well and in addition to that we also clarify several more concepts:

  • In-graph replication: the client creates a single tf.Graph that specifies tasks for devices on all workers. The client then creates a client session which will talk to the master service of a worker. Then the master will partition the graph and distribute the work to all participating workers.
  • Worker: A worker is a TensorFlow task that usually maps to one physical machine. We will have multiple workers with different task index. They all do similar things except for one worker checkpointing model variables, writing summaries, etc. in addition to its ordinary work.

The multi-worker version of this class maps one replica to one device on a worker. It mirrors all model variables on all replicas. For example, if you have two workers and each worker has 4 GPUs, it will create 8 copies of the model variables on these 8 GPUs. Then like in MirroredStrategy, each replica performs their computation with their own copy of variables unless in cross-replica model where variable or tensor reduction happens.

Args:

  • devices: a list of device strings.
  • num_gpus: number of GPUs. For local training, either specify devices or num_gpus. In distributed training, this must be specified as number of GPUs on each worker.
  • num_gpus_per_worker: number of GPUs per worker. This is the same as num_gpus and only one of num_gpus and num_gpus_per_worker can be specified.
  • cross_device_ops: optional, a descedant of CrossDeviceOps. If this is not set, the configure method will try to find the best one.
  • auto_shard_dataset: whether to auto-shard the dataset when there are multiple workers.
  • cross_tower_ops: Deprecated alias for cross_device_ops.

__init__

View source

__init__(
    devices=None,
    num_gpus=None,
    num_gpus_per_worker=None,
    cross_device_ops=None,
    auto_shard_dataset=False,
    cross_tower_ops=None
)

Properties

extended

tf.distribute.StrategyExtended with additional methods.

num_replicas_in_sync

Returns number of replicas over which gradients are aggregated.

Methods

experimental_distribute_dataset

View source

experimental_distribute_dataset(dataset)

Distributes a tf.data.Dataset instance provided via dataset.

In a multi-worker setting, we will first attempt to distribute the dataset by attempting to detect whether the dataset is being created out of ReaderDatasets (e.g. TFRecordDataset, TextLineDataset, etc.) and if so, attempting to shard the input files. Note that there has to be at least one input file per worker. If you have less than one input file per worker, we suggest that you should disable distributing your dataset using the method below.

If that attempt is unsuccessful (e.g. the dataset is created from a Dataset.range), we will shard the dataset evenly at the end by appending a .shard operation to the end of the processing pipeline. This will cause the entire preprocessing pipeline for all the data to be run on every worker, and each worker will do redundant work. We will print a warning if this method of sharding is selected.

You can disable dataset distribution using the auto_shard option in tf.data.experimental.DistributeOptions.

Within each host, we will also split the data among all the worker devices (if more than one a present), and this will happen even if multi-worker sharding is disabled using the method above.

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 distributed dataset
for x in dist_dataset:
  # process dataset elements
  strategy.experimental_run_v2(train_step, args=(x,))

Args:

  • dataset: tf.data.Dataset that will be sharded across all replicas using the rules stated above.

Returns:

A DistributedDataset which returns inputs for each step of the computation.

experimental_local_results

View source

experimental_local_results(value)

Returns the list of all local per-replica values contained in value.

Args:

  • value: A value returned by experimental_run(), experimental_run_v2(), extended.call_for_each_replica(), or a variable created in scope.

Returns:

A tuple of values contained in value. If value represents a single value, this returns (value,).

experimental_make_numpy_dataset

View source

experimental_make_numpy_dataset(
    numpy_input,
    session=None
)

Makes a dataset for input provided via a numpy array.

This avoids adding numpy_input as a large constant in the graph, and copies the data to the machine or machines that will be processing the input.

Args:

  • numpy_input: A nest of NumPy input arrays that will be distributed evenly across all replicas. Note that lists of Numpy arrays are stacked, as that is normal tf.data.Dataset behavior.
  • session: (TensorFlow v1.x graph execution only) A session used for initialization.

Returns:

A tf.data.Dataset representing numpy_input.

experimental_run

View source

experimental_run(
    fn,
    input_iterator=None
)

Runs ops in fn on each replica, with inputs from input_iterator.

DEPRECATED: This method is not available in TF 2.x. Please switch to using experimental_run_v2 instead.

When eager execution is enabled, executes ops specified by fn on each replica. Otherwise, builds a graph to execute the ops on each replica.

Each replica will take a single, different input from the inputs provided by one get_next call on the input iterator.

fn may call tf.distribute.get_replica_context() to access members such as replica_id_in_sync_group.

IMPORTANT: Depending on the tf.distribute.Strategy implementation being used, and whether eager execution is enabled, fn may be called one or more times (once for each replica).

Args:

  • fn: The function to run. The inputs to the function must match the outputs of input_iterator.get_next(). The output must be a tf.nest of Tensors.
  • input_iterator: (Optional) input iterator from which the inputs are taken.

Returns:

Merged return value of fn across replicas. The structure of the return value is the same as the return value from fn. Each element in the structure can either be PerReplica (if the values are unsynchronized), Mirrored (if the values are kept in sync), or Tensor (if running on a single replica).

experimental_run_v2

View source

experimental_run_v2(
    fn,
    args=(),
    kwargs=None
)

Runs ops in fn on each replica, with the given arguments.

When eager execution is enabled, executes ops specified by fn on each replica. Otherwise, builds a graph to execute the ops on each replica.

fn may call tf.distribute.get_replica_context() to access members such as replica_id_in_sync_group.

IMPORTANT: Depending on the tf.distribute.Strategy implementation being used, and whether eager execution is enabled, fn may be called one or more times (once for each replica).

Args:

  • fn: The function to run. The output must be a tf.nest of Tensors.
  • args: (Optional) Positional arguments to fn.
  • kwargs: (Optional) Keyword arguments to fn.

Returns:

Merged return value of fn across replicas. The structure of the return value is the same as the return value from fn. Each element in the structure can either be PerReplica (if the values are unsynchronized), Mirrored (if the values are kept in sync), or Tensor (if running on a single replica).

make_dataset_iterator

View source

make_dataset_iterator(dataset)

Makes an iterator for input provided via dataset.

NOTE: The batch size of the dataset argument is treated differently for this contrib version of MirroredStrategy.

Data from the given dataset will be distributed evenly across all the compute replicas. We will assume that the input dataset is batched by the per-replica batch size.

The user could also use make_input_fn_iterator if they want to customize which input is fed to which replica/worker etc.

Args:

  • dataset: tf.data.Dataset that will be distributed evenly across all replicas.

Returns:

An tf.distribute.InputIterator which returns inputs for each step of the computation. User should call initialize on the returned iterator.

make_input_fn_iterator

View source

make_input_fn_iterator(
    input_fn,
    replication_mode=tf.distribute.InputReplicationMode.PER_WORKER
)

Returns an iterator split across replicas created from an input function.

DEPRECATED: This method is not available in TF 2.x.

The input_fn should take an tf.distribute.InputContext object where information about batching and input sharding can be accessed:

def input_fn(input_context):
  batch_size = input_context.get_per_replica_batch_size(global_batch_size)
  d = tf.data.Dataset.from_tensors([[1.]]).repeat().batch(batch_size)
  return d.shard(input_context.num_input_pipelines,
                 input_context.input_pipeline_id)
with strategy.scope():
  iterator = strategy.make_input_fn_iterator(input_fn)
  replica_results = strategy.experimental_run(replica_fn, iterator)

The tf.data.Dataset returned by input_fn should have a per-replica batch size, which may be computed using input_context.get_per_replica_batch_size.

Args:

Returns:

An iterator object that should first be .initialize()-ed. It may then either be passed to strategy.experimental_run() or you can iterator.get_next() to get the next value to pass to strategy.extended.call_for_each_replica().

reduce

View source

reduce(
    reduce_op,
    value,
    axis=None
)

Reduce value across replicas.

Given a per-replica value returned by experimental_run_v2, say a per-example loss, the batch will be divided across all the replicas. This function allows you to aggregate across replicas and optionally also across batch elements. For example, if you have a global batch size of 8 and 2 replicas, values for examples [0, 1, 2, 3] will be on replica 0 and [4, 5, 6, 7] will be on replica 1. By default, reduce will just aggregate across replicas, returning [0+4, 1+5, 2+6, 3+7]. This is useful when each replica is computing a scalar or some other value that doesn't have a "batch" dimension (like a gradient). More often you will want to aggregate across the global batch, which you can get by specifying the batch dimension as the axis, typically axis=0. In this case it would return a scalar 0+1+2+3+4+5+6+7.

If there is a last partial batch, you will need to specify an axis so that the resulting shape is consistent across replicas. So if the last batch has size 6 and it is divided into [0, 1, 2, 3] and [4, 5], you would get a shape mismatch unless you specify axis=0. If you specify tf.distribute.ReduceOp.MEAN, using axis=0 will use the correct denominator of 6. Contrast this with computing reduce_mean to get a scalar value on each replica and this function to average those means, which will weigh some values 1/8 and others 1/4.

Args:

  • reduce_op: A tf.distribute.ReduceOp value specifying how values should be combined.
  • value: A "per replica" value, e.g. returned by experimental_run_v2 to be combined into a single tensor.
  • axis: Specifies the dimension to reduce along within each replica's tensor. Should typically be set to the batch dimension, or None to only reduce across replicas (e.g. if the tensor has no batch dimension).

Returns:

A Tensor.

scope

View source

scope()

Returns a context manager selecting this Strategy as current.

Inside a with strategy.scope(): code block, this thread will use a variable creator set by strategy, and will enter its "cross-replica context".

Returns:

A context manager.

update_config_proto

View source

update_config_proto(config_proto)

Returns a copy of config_proto modified for use with this strategy.

DEPRECATED: This method is not available in TF 2.x.

The updated config has something needed to run a strategy, e.g. configuration to run collective ops, or device filters to improve distributed training performance.

Args:

Returns:

The updated copy of the config_proto.