tf.contrib.distribute.MirroredStrategy

View source on GitHub

Mirrors vars to distribute across multiple devices and machines.

Inherits From: Strategy

tf.contrib.distribute.MirroredStrategy(
    devices=None, num_gpus=None, num_gpus_per_worker=None, cross_device_ops=None,
    auto_shard_dataset=False, cross_tower_ops=None
)

*** 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.

Methods

experimental_distribute_dataset

View source

experimental_distribute_dataset(
    dataset
)

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

The returned distributed dataset can be iterated over similar to how regular datasets can. NOTE: Currently, the user cannot add any more transformations to a distributed dataset.

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,))

We will assume that the input dataset is batched by the global batch size. With this assumption, we will make a best effort to divide each batch across all the replicas (one or more workers).

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. In this case, consider using experimental_distribute_datasets_from_function instead.

You can disable dataset sharding across workers using the auto_shard option in tf.data.experimental.DistributeOptions.

Within each worker, 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.

If the above batch splitting and dataset sharding logic is undesirable, please use experimental_distribute_datasets_from_function instead, which does not do any automatic splitting or sharding.

experimental_distribute_datasets_from_function

View source

experimental_distribute_datasets_from_function(
    dataset_fn
)

Distributes tf.data.Dataset instances created by calls to dataset_fn.

dataset_fn will be called once for each worker in the strategy. Each replica on that worker will dequeue one batch of inputs from the local Dataset (i.e. if a worker has two replicas, two batches will be dequeued from the Dataset every step).

This method can be used for several purposes. For example, where experimental_distribute_dataset is unable to shard the input files, this method might be used to manually shard the dataset (avoiding the slow fallback behavior in experimental_distribute_dataset). In cases where the dataset is infinite, this sharding can be done by creating dataset replicas that differ only in their random seed. experimental_distribute_dataset may also sometimes fail to split the batch across replicas on a worker. In that case, this method can be used where that limitation does not exist.

The dataset_fn should take an tf.distribute.InputContext instance where information about batching and input replication can be accessed:

def dataset_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)

inputs = strategy.experimental_distribute_datasets_from_function(dataset_fn)

for batch in inputs:
  replica_results = strategy.experimental_run_v2(replica_fn, args=(batch,))

experimental_local_results

View source

experimental_local_results(
    value
)

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

experimental_make_numpy_dataset

View source

experimental_make_numpy_dataset(
    numpy_input, session=None
)

Makes a tf.data.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.

Note that you will likely need to use tf.distribute.Strategy.experimental_distribute_dataset with the returned dataset to further distribute it with the strategy.

Example:

numpy_input = np.ones([10], dtype=np.float32)
dataset = strategy.experimental_make_numpy_dataset(numpy_input)
dist_dataset = strategy.experimental_distribute_dataset(dataset)

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.

experimental_run_v2

View source

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

Run fn on each replica, with the given arguments.

Executes ops specified by fn on each replica. If args or kwargs have "per-replica" values, such as those produced by a "distributed Dataset", when fn is executed on a particular replica, it will be executed with the component of those "per-replica" values that correspond to that replica.

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

All arguments in args or kwargs should either be nest of tensors or per-replica objects containing tensors or composite tensors.

make_dataset_iterator

View source

make_dataset_iterator(
    dataset
)

Makes an iterator for input provided via dataset.

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.

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.

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.

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".

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.